Downloaded from orbit.dtu.dk on: Oct 09, 2021
High-intensity, focused ultrasonic fields
Jensen, Leif Bjørnø
Published in: Acoustical Society of America. Journal
Link to article, DOI: 10.1121/1.2025494
Publication date: 1988
Document Version Publisher's PDF, also known as Version of record
Link back to DTU Orbit
Citation (APA): Jensen, L. B. (1988). High-intensity, focused ultrasonic fields. Acoustical Society of America. Journal, 83(S1), S71-S71. https://doi.org/10.1121/1.2025494
General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.
Users may download and print one copy of any publication from the public portal for the purpose of private study or research. You may not further distribute the material or use it for any profit-making activity or commercial gain You may freely distribute the URL identifying the publication in the public portal
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
PROGRAM OF
The 115thMeeting of the AcousticalSociety of America
SeattleSheraton Hotel andTowers ß Seattle,Washington ß 16-20 May 1988
MONDAY EVENING, 16 MAY 1988 WEST BALLROOM, 7:00 TO 9:00 P.M.
Tutorial Lecture
Acousticintensity technique. Jiri Tichy(Pennsylvania State University, Graduate Program in Acoustics, AppliedResearch Laboratory, University Park, PA 16802) Radiationand propagation ofsound energy represents fundamental knowledge in all subfields ofacoustics andvibration. The intensity technique was initially developed for sound power measurements butit canbe used asa generaltool for sound field investigations. This is possible by recent developments ofsignal processing and precisioninstrumentation technology which permit identification of detailed behavior of vibratingsurfaces, mappingof soundfields in termsof soundpressure, particle velocity, complex acoustic intensity and energy densities,construction ofwave fronts, and determination of phaseand group velocities. The tutorial will cover thefundamentals of all aspectsof theintensity technique. Basic theoretical relationships of sound energy generationand propagation will be discussed. The basics of the measurement technique, using two-microphone probesor pressure-velocityprobes, and the appropriate dedicated instrumentation or more general computer use,will be discussed. The tutorial will briefly handle the precision ofthe measurements. A great deal of time willbe devoted to examplesof applicationson radiation from simple and complex sources and their nearfields. Themaps of energy propagation, sound pressure and particle velocity, as well as the formation of vortices,will beshown. Specific applications in manyareas of acousticswill besummarized. Afterthe lecture, various companies will demonstratethe latest commercial equipment available to make acousticintensity measurements.
TUESDAY MORNING, 17 MAY 1988 GRAND BALLROOM C, 8:15 TO 11:42 A.M.
SessionA. SpeechCommunication I: Analysis, Coding, and Synthesis
Peter Benson, Chairman I TTDefence Communication Division, 10060 Carroll Canyon Road, San Diego, California 92131
Chairman's Introduction--8:15
ContributedPapers
8:17 8:32
A1. An optimal classof generalizedtime-frequency representations for A2. Formant estimation from noisy voiced speech. A. nonstationarysignal analysis. Yunxin Zhao, Les E. Atlas, and Robert K. Krishnamurthy,J. Li, and R. L. Moses(Department of Electrical J. Marks, II (Department of ElectricalEngineering, FT-10, University Engineering,The Ohio StateUniversity, 205 DreeseLaboratories, 2015 of Wnahington,Seattle, WA 98195) Neil Avenue, Columbus, OH 43210)
In this paper,a classof generalizedtime-frequency representations This paperconsiders the estimation for formantfrequencies and band- (GTFR) that haveboth good time and good frequency characteristics for widthsfrom voicedspeech signals that aredegraded with additivewhite nonstationarysignal analysis is presented.The basisof the approachused noise.Each period of thespeech signal is dividedinto theopen and closed isthat a time-frequencyrepresentation can, in somecases, be improved by glottalphases using the electroglottograph signal. The speech signal in the allowingnegative value of thespectrum. The representation(and its asso- closedphase is modeled as the sum of dampedsinusoids, with eachsinus- ciatedkernel) can then be optimizedto enhancethe peaksin frequency oldcorresponding to oneformant. The analysisprocedure thus leads to an whilestill maintainingfinite time support. The finitetime support proper- estimateof the frequency,bandwidth, and energyof eachformant. The ty hasthe significance of preservingonset times of signalsand also provid- resultsindicate that typicallyjust threeformants account for mostof the ing clear representationsof fast-changingspectral peaks. Experiments energyin the closedphase. The noise-robustnessproperties of the algo- wereperformed on simulateddata and real speechfor comparisonof the rithm are increasedby usinga total least-squaresapproach in obtaining GTFR with the spectrogramand the pseudo-Wignerdistribution. The the parametersand by using data from multiple consecutiveclosed resultsshow distinctadvantages of the GTFR. For example, onsetsof phases.This work verifiesand extendsthe resultsof S. Parthasarathyand transitionsare clearerand, in the caseof speech,close formant peaks are D. Tufts [IEEE Trans.Acoust. Speech Signal Process. ASSP-35, 1241- easierto distinguish.[Research supported by NSF and Boeing.] 1249(1987) ]. [Work supportedby theOSU Seedgrant program. ]
$1 d. Acoust.Soc. Am. Suppl. 1, Vol. 83, Spring 1988 115th Meeting:Acoustical Society of America $1 8:47 9:17
A3. Voiced speechmodel includingsource-tract interaction. A. K. AS. A real-time frequency-shift decoder for the last transmissionof Krishnamurthyand J. Li (Departmentof ElectricalEngineering, The Korean Airlines flight 007. John D. Schlatter and Les E. Atlas Ohio State University, 205 Dreese Laboratories,2015 Nell Avenue, (Department of Electrical Engineering, FT-10, University of Columbus, OH 43210) Washington,Seattle, WA 98195)
This paperconsiders the problemof estimatingthe parametersof a On 1 September1983, Korean Airlinesflight 007 wasintercepted and voicedspeech model that includesthe effectsof source-tractinteraction. destroyedby a Sovietjet fighterafter strayinginto Sovietairspace. Al- Basedon work by Ananthapadmanabhaand Fant [ SpeechCommun. 1, though the flight recorderfrom KAL 007 was not recovered,its final 167-184 (1982) ], the voicedspeech signal is modeledas the outputof a transmissionswere recordedat a Tokyo air traffic control center. The time-varyingvocal tract filter excitedby a parametrizedglottal source intelligibilityof the KAL 007 tansmissionsis compromisedby at least waveform.The glottalsource waveform [ Fant et al., STL/QPSR (1984) ] three types of distortion:broadband background noise, narrow-band modelsthe main pulse shapeof the glottal volume velocity and is de- noisetone, and frequency-shiftdistortion. This paper describesa DSP scribedby four parameters.The effectof source-tractinteraction is mod- systemthat hasbeen developed to counteractthe frequency-shiftdistor- eledby varyingthe frequency and bandwidth of thefirst formant in synch- tion. The systememploys Hilbcrt transform-basedtechniques to intro- rony with the glottal source waveform. An analysis-by-synthesis ducea linearcompensating frequency shift, the extentof whichmay be approachis usedto estimatethe parametersof the model.Algorithms for controlledmanually by theuser or automaticallyby a separatecomputer. estimatingthe parametersof the glottalsource waveform and the vocal Testsperformed to date with the systemon the KAL 007 transmissions tractfilter are described. Comparisons of thespectrum of theoriginal and haveshown it to bea usefultool. Intelligibility improvements have led to synthesizedspeech waveforms are presented.[Work supportedby the independentlyverified advances in understandingseveral of KAL 007's OSU Seedgrant program. ] final transmissions.
9:32 9:02 A6. Large-band integrated vowel representations.Marlos Mantakas, A4. Distribution of spectral errors with quantization of frequenciesand Jean-Luc Schwartz, and Pierre Escudier (Laboratoire de la bandwidths of LPC poles. Bishnu S. Atal (Acoustics Research Communication Par16e, I.C.P.-Unit6 associ6e au C.N.R.S., Department,AT&T Bell Laboratories,Murray Hill, NJ 07974) E.N.S.E.R.G.-I.N.P.G., 46 avenue F61ix-Viallet, 38031 Grenoble, France) In lowbit ratespeech coding systems, the LPC spectralinformation is oftencoded in termsof quantizedpartial correlations or line-spectral "Gross"vowel spectrum parameters for vowelclassification are of pairs.The distributionof spectralerrors with suchquantization has been interestto manyresearchers. In thepresent study, the "effectiveformant" studiedin the past.However, not muchis knownabout the distribution of F'2, estimatedby thelarge-band spectral integration (LBI) model[ Escu- spectralerrors when the frequenciesand bandwidthsof LPC polesare diet et aL, Acts of the French-SwedishSeminar, Grenoble, France quantized.The quantizationof frequenciesand bandwidthshas impor- (1985) ], in the classificationof natural Frenchfront vowels/i/vs/y/ tant advantagesbecause the perceptual sensitivity to quantizationof these and/e/vs/•/(rounding opposition), isevaluated. In parallel,formants' parameterscan be estimatedfrom data onjust-noticeable differences. In measureson the samecorpus are providedin orderto comparewith and thispaper, results will bepresented on thespectral error introduced by the interpretthe resultsobtained using the LBI model Classificationperfor- quantizationof frequenciesand bandwidths of LPC polesas a functionof manceis good in the caseof/i/-/y/. The optimal spectralintegration the differentnumber of predictorcoefficients and the differentnumber of window is 2-2.5 Bark large (cf. with the 3-3.5 Bark "critical distance"). bits usedto representthe spectrum.The distributionof spectralerrors The model fails, however,in the/e/-/•/classification. Here, the second usingfrequency-bandwidth quantization will be comparedwith other formant'sfrequency is by far winning.The model'sweakness apparently proceduresbased on quantizationof partial correlationsand line-spectral residesin its last stage,namely peak estimation.In fact, recentresults pairs, and relative strengthsand weaknessesof these quantization showthat the LBI spectralrepresentation can supportthe definitionof schemes will be discussed. form factors(other than peakposition) for the classificationof vowels.
9:47-9:57
Break
9:57 bilitytesting using target speech ofread PB sentences masking read four- word nonsense sentences revealed no differences between treated and un- A7. On the semanticcontent of intelliglbility test material on assessing treated speech.Intelligibility usingcomplete sentences as both masker signalprocessing algorithms. P. Bensonand J. Naylot (ITT Defense and target showedsignificant differences in favor of the HMS treated CommunicationDivision, San Diego, CA 92131) speech.An analysisof signalsand errorswill be presentedto showhow Signalprocessing algorithms can be usedto enhancespeech intelligi- listenersare integrating information from nonsignal sources to improve bility by improving the detectabilityof important acousticfeatures. It performance. shouldfollow from this that othernonacoustic portions of the signalare a sourceof experimentalerror that shouldbe controlled.Nevertheless, cer- tain signalprocessing techniques produce reliable improvements in intcl- ligibility only when that speechis meaningful.A seriesof intelligibility 10:12 testswas carried out with co-channelspeech, which combines the speech of two talkersinto a singlechannel. This speech had been processed using AS. Preliminary study of multilevelpeak-clipped and time-quantized the harmonicmagnitude suppression (HMS) technique[Naylor and speech.Edward M. O'Brien and David J. Poguc (Bioengineering Boll, ICASSP (1987) ] to diminishthe effect of thelouder talker. Intclligi- Program,Texas A&M University,College Station, TX 77843-3120)
$2 J. Acoust.Soc. Am. Suppl. 1, Vol. 83, Spring 1988 115thMeeting: Acoustical Society of America S2 Speechwas amplified to a levelof plusand minus 5 V foranalog-to- 10:57 digitalconversion (sampling rate of 20 kHz). It wasthen quantized to five discretelevels. The 0-V levelwas used for speech within the squelch level All. Delta streamsfor English synthesis.Susan R. Hertz (Eloquent (squelchwas used to maskbackground noise). Positive speech waves Technology,254 Highgate Circle, Ithaca, NY 14850 and Phonetics aboveA voltswere assigned a value of 5 V. Speechwaves with peak values Laboratory,Cornell University,Ithaca, NY 14853) belowthis level were assigned a value of A volts.Negative speech waves Although speechpatterns are heavily influencedby the hierarchical werequantized to either -- 5 V or - A voltssimilar to theprocessing of linguisticstructure of speech,synthesis by rule systemshave generally thepositive speech waveforms. Processing was done with a PC. Two dif- beenbased on linearutterance representations. Delta isa newprograming ferentA levelswere tested on monosyllabicword lists.The firstwas at the languagethat makesit easyto work with utterancerepresentations con- levelof the rmsvalue of eachword (averagerms levelacross all wordswas tainingmultiple levels. Both higher level linguistic units, such as phrases, 1.22V). The secondlevel was at a fixedvalue of 2 V. Preliminaryresults syllables,and phonemes, and lower level phonetic events, such as articula- yieldeda higherpercentage word intelligibility. This type of speechpro- tory or formanttargets and F0 trends,can be easilyaccommodated on cessinghas application to the developmentof a tactilehearing aid. separateinterconnected "streams," with eachunit equallyaccessible to therules. While Delta canbe used to testmost any synthesis model for any language,this paper will showhow Delta canbe usedto testa particular modelfor English.This modeluses, among others, CV, syllable,nucleus, phoneme,formant, and duration streams, with formanttransitions repre- sented as duration tokens that are in effect invisible in other streams. The paperwill justifythe.selection of streamsand the uniqueway of handling 10:27 formanttransitions, demonstrating in Delta notationhow the modelleads A9. A microcomputer-basedsystem for high-speedprocessing of speech to particularlystraightforward rules for predictingEnglish phoneme du- materialsusing pitch-synchronous LPC analysisand synthesis.Stephen rations,formant values,and aspirationpatterns. J. Eady, B. Craig Dickson, Roy C. Snell (Centre for Speech TechnologyResearch, University of Victoria,P.O. Box 1700,Victoria, BritishColumbia V8W 2Y2, Canada),and MelvynR. Hunt (National ResearchCouncil of Canada,Building U-61, Montreal Road, Ottawa, Ontario K 1A 0R6, Canada) 11:12
Thispaper describes a microcomputer-based system that uses a high- AI2. A new designof a digital conferencesystem. Hans W. Gierlich speeddigital signal processor (TMS-32030) for digitizingand encoding (Institute for CommunicationEngineering, RWTH Aachen,Melatener speechmaterials. The systemuses pitch-synchronous LPC (covariance Strasse23, 5100 Aachen,West Ocrmany) method),in conjunctionwith a laryngograph,to analyzepreviously re- Two main drawbacksmake telephoningusing a conferencesystem cordedspeech materials [M. J. Hunt andC. E. Harvenberg,Proceedings troublesome:First, noiseand reverberationproduced at the site of the ofthe 12th International Congress on•4 coustics, A4-2 ( 1986) ]. Thespeech speakerare transmittedto the listener.Second, to achievea sufficient andlaryngograph signals are digitized simultaneously, and speech param- speechvolume at the listener'ssite, the necessaryamplification for loud- eters(F0, energy,formants, and bandwidths)are calculatedin real time. speakerplayback leads to instability(feedback) of the wholesystem. To The parametersare modifiedusing an interactiveeditor, and real-time suppressreverberation and noise,a speechcontrolled microphone ar- synthesisallows for auditorymonitoring of the originaland modified rangementis used.The implementedalgorithm for speechdetection is speechsamples. The systemis currentlybeing used to processspeech ableto distinguishbetween noise and speechup to noiselevels of 70 dB materialsfor a limited-vocabulary,word-concatenation synthesis system SPL. The noisesignal may either be noiselike,impulselike, or sinusoidal. [Eadyet al., Proc. I CASSP,1473-1476 (1987) ], andalso in thedevelop- The detectionalgorithm needs a maximumtime of 32 ms to detecta mentof a demisyllable-basedtext-to-speech device. [Work supported by speechsignal. To avoid instability,a combinationof a specialfilter ar- National Research Council Canada and Science Council of British Co- rangementand a speechcontrolled amplification of microphoneand lumbia. ] loudspeakersignal is used.Two filters, having a transfercharacteristic similarto a combfilter, are insertedinto the transmittingand the receiv- ing path. The transferfunctions of the filtersare inverse.With regardto the tone color, the filterswere optimized, taking into accountthe special effectsof signalprocessing used by the humanear. Hearingtests showed that speechintelligibility is not influencedby the filter arrangement. 10:42 A10.Adjusting syllable durations in'a demisyllablespeech synthesis system.Suzanne C. Urbanczyk,Stephen J. Eady,and B. Craig Dickson (Centre for SpeechTechnology Research, University of Victoria, P.O. 11:27 Box 1700, Victoria, British Columbia V8W 2Y2, Canada) A13. The smoothedpseudo-Wigner distribution in speechanalysis. Labib This paperdescribes the durationalaspects of a speechsynthesis sys- M. Khadra (The University of Trondheim, Division of temdesigned togenerate words and sentences ofEnglish based on smaller Telecommunications,Trondheim, Norway) syntheticspeech units known as demisyllables [O. Fujimura,J. Acoust. Soc.Am. Suppl.159, S55 (1976) ]. Eachsyllable of a wordis composed of Recently,a great deal of interesthas beenshown in applyingthe an initial and a finaldemisyllable. The demisyllableinventory (totaling Wignet-Villedistribution to obtaintime-frequency energy representation lessthan 900 units)is prerecorded and stored in LPC format.Demisylla- of nonstationarysignals. The utilityof theWigner distribution has proved blesare linked to formsyllables, which are then joined into words. Syllable usefulin analyzingmonocomponent signals. In the caseof multicompo- durationisone of the more complicated features of wordformation using nentsignals, the Wignetdistribution adds cross terms without any phys- thistechnique. Our strategyfor producingsyllables of appropriatelength icalsignificance to the time-frequencydistribution. The presenceof cross isto recordthe demisyllableunits with relativelylong durations and then termsobscures the actual spectralfeatures of interestand makesthe re- to reducethe durationof the demisyllablecomponents when they are sultsvery misleading.To applythe Wignet-Ville distributionto speech joinedtogether. Durational adjustment is accomplishedusing a spectral signals,it is essentialto removeall crossterms not of interest.This prob- distancemetric, which identifies regions of the syllablewhere the spectral lemmay be solved by smoothing the Wigher-Ville distribution indepen- componentsare least dynamic. The less dynamic regions can then be dently in time and frequencydirections using the "smoothed"pseudo- deletedto shortena syllable,while maintaining perceptually relevant Wigher estimator.This estimatorpossesses several advantages over the acousticcharacteristics. This method is usedto adjustsyllable durations shorttime periodogram.This presentationwill concentrateon the appli- at both the word and sentencelevels. [ Work supportedby ScienceCoun- cationof the smoothedpseudo-Wigner distribution (SPWD) to speech cil of British Columbia.] signalsand will demonstratethe abilityof the SPWD to improvethe
$3 J. Acoust.Soc. Am. Suppl. 1, Vol. 83, Spdng 1988 115th Meeting:Acoustical Society of America S3 qualityof thetime-frequency representation ofspeech signals. In particu- detectedfrom an SPWD but are far lessobvious in the spectrogram. lar,a comparisonwill be made between the spectrogram and the SPWD, Moreover,it will be shownthat the SPWD is more appropriatefor the andit will beshown how high-frequency spectral features may be easily analysisof formantstructure.
TUESDAY MORNING, 17 MAY 1988 WEST BALLROOM A, 8:30 TO 11:45 A.M.
SessionB. Physical AcousticsI: Nonlinear Acoustics
JacquelineNaze Tj½ttta and Sigve Tjrtta, Cochairmen AppliedResearch Laboratories, The Unioersity of Texasat Austin,/lustin,Texas 78713-8029 and Department of Mathematics,The Universityof Bergen,5007 Bergen,Norway
ContributedPapers
8:30 earlier solutionfor finite amplitudesound beams [J. H. Ginsberg,H. C. Miao, andM. A. Foda,J. Aconst.Soc. Am. Suppl.181, S25 ( 1987) ]. The B1. Ray solutionfor finite amplitudetwo-dimensional waves in a hard- signalin thatanalysis was represented in termsof a spectrumof transverse walled rectangular waveguide. Kun-tien Shu and J. H. Ginsberg wavenumbers,rather than frequencies,and the signalfor eachwavenum- (Schoolof MechanicalEngineering, Georgia Institute of Technology, ber wasformed from interacting quasiconical waves, rather than a single Atlanta, GA 30332) planarwave. Nevertheless, the two problems share a commonmathemat- This paperdescribes a ray superpositiontheory for cumulativegrowth icalstructure. A discussionof numericalalgorithms for such models leads of nonlinear effectsin a two-dimensional acoustic mode, based on decom- to somesurprising observations regarding the implicitfunctional form of positionof themode into a pairof obliquelypropagating, nonlinear planar theEarnshaw solution. [Work supportedby ONR.] waves.The mathematicalfoundation of the formulationis an earlierper- turbationanalysis of the reflectionof a distortedplanar wave obliquely incidenton the boundaryof an infinitehalf-space [ Z. Qian, Sci. Sin. 25, 9:00 492-501 (1982) ]. Basedon the •sults of the analysis,each of the pair of BJ. Scatteringof soundby soundfrom two real beams.Jarle Berntsen raysforming the signalat a selectedfield point is tracedback to its origin (Department of Infotreaties,The Universityof Bergen,5007 Bergen, at the excitation.Each ray is describedas a simpleplanar wave undergo- Norway), JacquelineNaze Tj•tta, andSigve Tj•tta (AppliedResearch ing finite'amplitude distortion that dependson the propagationdistance Laboratories,The Universityof Texasat Austin,Austin, TX 78713-8029 alongthat ray betweenfield and sourcepoints. This distanceis the same andDepartment of Mathematics,The Universityof Bergen,5007 Bergen, for eachray at a specifiedfield point, but differencesin the excitationat Norway) the respectivesource points result in phasedifferences between the two rays.The overallsignal is shownto be the sameas a modaldescription of A theory for the nonlinearinteraction between two soundbeams pro- thepropagation [H. C. Miao and$. H. Ginsberg,I. Acoust.Soc. Am. 80, ducedby realsources in a losslessfluid was presented in a previouswork 911-920 (1986) ]. The ray solutionexplains an apparentparadox in the [Naze Tjrtta and Tjrtta, J. Acoust.SOC. Am. 83, 487-495 (1988)]. A modalanalysis, which indicated that althoughthe signalcan be resolved generalsolution of the governingequation in the quasilinearapproxima- into a pair of planarwaves, the distortionprocess is scaledonly by the tion,valid at anyrange, crossing angle, and frequency ratio, was obtained axial positionalong the waveguide.Conversely, the earliersolution pro- for prescribedboundary conditions. An asymptoticformula for the sum videsvalidation for the superpositionof rays,as well as for the linear and differencefrequency sound pressure was obtained at largedistance reflectionlaw. [Work supportedby NSF and ONR.] fromthe sources. It relatesthe amplitude and directivity of thesound field in thefarfield to theon-source conditions. In thepresent work, numerical resultsare presentedfor varioustypes of sources(uniform piston,Gaus- siansource, focusing soureeL The influenceof sourcegeometry (separa- tion, and intersectionangles from 0• and90 •} and frequencyon the beam patternof thenonlinearly generated sound is studied. Conditions are also 8:45 givento determinewhen scattering of soundby soundcan be observed. In B2. Similarity of a Fourier trimsformgeneralization of the Earnshaw the specialcase of thin Gaussianbeams intersecting at a smallangle, the solution for planar waves to an interacting wave model for finite resultsare compared with that presented by Darvennesand Hamilton [J. Acoust.Soc. Am. Suppl.1 83, S4 (1988) ] usingthe paraxialapproxima- amplitudeeffegts in soundbeams. J. H. Ginsberg(School of Mechanical tion. [Work supportedby the IR&D programof ARL:UT, andVISTA/ Engineering,Georgia Institute of Technology,Atlanta, GA 30332) STRATOIL, Norway. ] The Earnshawsolution for a finiteamplitude planar wave, which dis- playsamplitude dispersion, is valid for arbitraryexcitation f(t) on a boundary.The formthat results for smallacoustic Mach numbers when f(t) is representedas the inverse of its Fouriertransform F(co) may be consideredto be a coordinatestraining of the linearizedsignal, in which 9:15 thetransformation has the appearanc• of a Fredholmintegral equation in B4. Scatteringof soundby soundfrom two Gaussianbeams. Corinne thefrequency domain. In thecase of an acousticplanar wave, the phase M. Darvennesand Mark F. Hamilton (Departmentof Mechanical speedof all frequencycomponents isthe same, independent offrequency. Engineering,The University of Texas at Austin,Austin, TX 78712-1063) A generalizationof thatresult is obtained if oneconsiders the possibility that the cumulativegrowth effect is an arbitraryfunction of distance. The scatteringof soundby soundfrom Gaussian beams that interact Ratherthan being an abstraction, this form is shown to beanalogous to an at smallangles is investigatedtheoretically with a quasilinearsolution of
•4 J.AcousL Soc. Am. Suppl. 1,Vol. 83, Spring 1988 115thMeeting: Acoustical Society ofAmedca •4 theKhokhlov-Zabolotskaya nonlinear parabolic wave equation. The an- 10:00 alyticalsolution,.which is valid throughout the entire paraxial field, is a generalizationof the result obtained for parametricreceiving arrays by 87. Weakfoci in three-dimensionallinear shock waves and the stability of Hamilton,Naze Tj•Jtta,and Tj•tta [J. Acoust.Soc. Am. 82, 311-318 real shockwaves. P. L. Marston(Department of Physics,Washington (1987)]. Significantlevels of scattereddifference frequency sound are StateUniversity, Pullman, WA 99164) shownto existoutside the nonlinear interaction region. An asymptotic A firstapproximation for identifyingfoci in weakshocks is to consider formulareveals that difference frequency sound is scatteredin theap- thefocusing process for a lineardiscontinuity. This is sometimes known as proximatedirection of k•-k:, whereki is thewave vector associated with the"acoustic shock" approximation. A case studied previously isthe lon- thedirection and frequency ofthe ith primary beam. Computed propaga- gitudinalcusp of"ar&c" locatedwhere three rays merge at thefocus of a tion curvesand beampatterns demonstrate the dependenceof the scat- convergingcylindrical wave; for finiteamplitudes, the wavecan become teredradiation on sourceseparation, frequency ratio, interaction angle, unstablewith respectto the formationof a pair of shock-shocksin the andfocusing. Results are also presented for the scattered sum frequency focalregion lB. Sturtevantand V. Kulkarny,J. FluidMech. 73, 651-671 sound.Comparisons are madewith the generalasymptotic results pre- (1976) ]. In the presentresearch, catastrophe theory is usedto identify sentedby Berntsen, Naze Tj4tta, and Tj4tta [J. Acoust. Soc. Am. Suppl. 1 wave-frontshapes W(x,y) that propagateto producenovel acoustic foci 83, S4 (1988)], whichare valid for arbitraryinteraction angles, source with causticsspread out roughlytransverse to the directionof propaga- separations,and amplitude distributions. [Work supportedby ONR.] tion [P. L. Marston, J. Acoust. Soc.Am. 81, 226-232 (1987) and Pro- ceedingsof the APS 1987 TopicalConference on Shock Wavesin Con- densedMatter(in press)]. The initialdisplacement of thewave front from the xy plane is givenby W(x,y). Linear waveswith W= -- (a,x• + az•x + any•), wherethe shape parameter a:• 0,propagate toproduce 9:30 transversecusps described by a cubiccusp curve. A strongerfocus, the hyperbolic umbilic, is produced if W= -- [(OtX3 + 3yy2X)/ 85. Acoustical phase conjugationexperiments: The generationof a 6 + a, (x: + y2)] with a, <0. A focalsection is producedat a distance reversed wave through three-wave mixing in a layer of stabilized z = -- (2aI ) - • fromthe xy plane.In thissection, caustic lines form a V microbubbles.Steven G. Kargl and P. L. Marston (Department of withan apex angle of 2 arctan (7'/a) •/2Issues related to shock stability are Physics,Washington State University, Pullman, WA 99164-2814) noted.[Work supportedby ONR.] A phase-conjugatemirror is onethat reversesan incidentwave front sothat it propagatesback toward the source.Recent experiments [ Kus- toy et al., Soy.Phys. Acoust. 32, 500-504 (1986) ] indicatethat wave- frontreversal can be established through the interactionof a pumpwave 10:15 of frequencyf• with a probewave of frequency.f:
$5 d. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115th Meeting:Acoustical Society of America S5 order of approximation.It is shownthat secularsolutens, which are thanthe simple reduction of visconsshear losses. [Work supported by the usuallyassociated with cumulativelygrowing nonlinear distortion, can Officeof Naval Researchand the Officeof Naval Technology.] alsobe producedby the inhomogeneity,if the periodicityof the exciting stressand inhomogeneity satisfy a certainrelation. [Work supportedby NSF. ] 11:15
B12. The effect of mean flow gradients, including shear, on the propagation of sonnd. P.O. Vaidya (Mechanical and Materials 10:45 EngineeringDepartment, Washington State University,Pullman, WA 99164) BI0. An investigationof the work output of a thermoacousticprime mover.Daniel C. Simard,Anthony A. Atchley, and StevenR. Baker In the caseof the perturbationanalysis of nonlinearacoustics, it is (Departmentof Physics,Naval PostgraduateSchool, Monterey, CA customaryto assumethat the underlyingundisturbed mean flow is uni- 93943) form or nonexistent.When this assumption is removed,several interest- ingphenomena occur, which are of considerablepractical significance. In A thermoaconsticprime moveris a devicethat exploitsan imposed thispaper, three kinds of meanflow nonuniformities are analyzed. These temperaturegradient to generatea workoutput. This workserves to offset areshear flows, axial gradients, and mean flows with nonzero divergence. dissipativelosses and thus improves the quality factor of oscillations.The It hasbeen shown that these cases lead to soundattenuation, amplifica- theoryis developedusing a quantitativeexpression for thermoacoustic tion,and instabilities, as well as the generation of fluctuating vorticity and work as derivedby Swift et al. at Los AlamosNational Laboratory [J. the acousto-vortical waves. Aeoust.Sec. Am. 78, 767-776 (1985)]. An experimentis describedin whichthe quality factor is measuredas a functionof temperaturegradi- ent.A modifiedversion of the "Hofler"tube is usedand the experiment is conductedin argon.The resultsindicate that the inverseof the quality 11:30 factoris linear with temperaturegradient and are in goodagreement with theory. [Work supportedby ONR.] BI3. Effectof stresson the temperaturedependence of soundvelocity. JohnH. Canttell(NASA LangleyResearch Center, Mail Stop231, Hampton,VA 23665-5225) The acousticnatural velocity for crystallinesolids under stress is ex- 11:00 presseddirectly as a functionof the stressand the second-and third-order B11. Low Prandtl number ga• mixtures as u working fluid in a elasticconstants referred to the zero stress state. The temperature depend- thermoacousticrefrigerator. M. Suzalla, T. Hofier, and S. L. Oarrett enceof theelastic constants ismade explicit by use of theHelmholtz free (Departmentof Physics,Naval PostgraduateSchool, Monterey, CA energyfor a systemof quantizedharmonic oscillators. The resulting 93943) expressionis usedto calculatethe temperaturederivatives of the sound velocity(dv/dT) asa functionof appliedstress a in thesolid. It isfound Prandtlnumber (Pr) isthe dimensionless ratio of kinematicviscosity that(dv/dT )o- • [ (dv/dT)o - (dv/dT)o] = Ktr,where the subscript- tothermal diffnsivity, and is about 0.7 for most ideal gases. This value can edtr denotesevaluation at stresstr andthe subscripted zero denotes eva- be loweredsignificantly by mixingtwo gasspecies having molecular luationat zerostress. The constant K isfound to dependexplicitly on the weightsthat are very different, resulting in a minimumPr of 0.22for He- third-,fourth-, and fifth-order elastic constants of thecrystal. A central- Xe mixtures.This can be usedto minimize the nuisanceeffect of viscous forcenearest-neighbor lattice model is usedto calculatethe fourth-and shearlosses for a thermoacousticrefrigerator aswell as for other types of fifth-orderconstants from measured values of theacoustic nonlinearity heatengines. The principle of thermoacousticheat transport will be brief- parameters.For these elastic constant data, K (forstress applied perpen- ly discussed[J. WheatIcy,T. Hofier,G. W. Swift,and A. Migliori,J. dicularto the longitudinal wave propagation direction inisotropic solids) Acoust.Sec. Am. 74, 153-170( 1983) ]. However,changing the viscosity iscalculated to be 1.8X 10-3 MPa- t for purealuminum and 6.9 X 10-4 ofthe working fluid also changes the details of the acoustic velocity distri- MPa-• for copper. These results are in good agreement with experimental bution,thereby modifying the basic thermoacoustic heat transport mech- valuesobtained for aluminumand copper alloys [K. Salamaand C. K. anism.Measurements indicate that thiseffect may be moreimportant Ling,J. Appl. Phys.51, 1505(1980)].
TUESDAY MORNING, 17 MAY 1988 EAST BALLROOM A, 8:30 TO 11:35 A.M.
SessionC. StructuralAcoustics and Vibration I: SpecialSession on Response of CoupledDynamic Systems
Gideon Maidanik, Chairman DavM TaylorResearch Center, Code 1902, Bethesda, Maryland 20084
Chnirmnn's Introduction---8:30
Invited Papers
8:40
C1. Modelingstructural acoustics for activecontrol. A. H. von Flotow, D. W. Miller, and D. J. Pines (Departmentof Aeronauticsand Astronautics,Massachusetts Institute of Technology,Cambridge, MA 02139) Activetechniques are beginning tobe considered forthe control of structure-borne sound. Expertise from manyfields is being brought to bearupon this problem, including experience in activecontrol of theacoustic
$6 J. Acoust.Sec. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S6 responseof ductsand air volumes,and the activecontrol of flexiblespacecraft. High-fidelity modeling is a necessaryprerequisite to high-performanceactive control. This paperbriefly reviews the pastapproaches to themodeling of structure-bornesound and approaches taken for theirverification. The mainbody of the paper explainsa recentlydeveloped approach to suchmodeling, applicable to a restrictedclass of structures.The papercloses with the descriptionof two techniquesdeveloped for the designof activecontrol systems for structure-bornesound, and describespreliminary laboratory experimental efforts.
9:05
C2. Responseofcoupledone-dimensionaldynamiesystems.J. Dickey, L.J. Maga,andG. Maidanik (David Taylor ResearchCenter, Annapolis,MD 21402) Attentionis focusedon a complexof one-dimensionaldynamic systems. The extentera dynamicsystem is definedin termsof two terminalpositions, one at eachof twojunctions. A terminalvector per junction can thus be constructed.The propagationin a dynamicsystem is definedin termsof two propagationfunctions each describingthe propagationtoward one or theother junction. A diagonalpropagation matrix per junction can thusbe constructed. The couplingamong the dynamic systems is defined in termsoftwe junctionmatrices. The externaldrive at a positionon a givendynamic system is defined in termsof theresponses that it initiatestoward onejunction and towardthe otherjunction. With thismodel and definitionsof a complexof coupledone- dimensionaldynamic systems one may appropriately derive the variouscomponent matrices of the impulse responsematrix. The matrix yieldsthe responseof the complexto the appropriatelydefined external drive vector.The derivationis explainedand discussed. The basic(exponential-type) propagation forms and the nonbasic(e.g., Bessel functional-type) propagation forms are contrasted. The utilityof theformalism is briefly cited and discussed.
9:30
C3. Structural power flow analysisof coupledstructures. J. M. Cuschieri(Center for Acousticsand Vibration,Department of OceanEngineering, Florida Atlantic University, Boca Raton, FL 33431) Structuralpower flow techniques are an effective tool for thedynamic analysis of complexcoupled struc- tures,especially those types of coupled structures that are somewhat repetitive. Using the structural power flow method,the responseof a structureis analyzedby trackingthe flowof the vibrationalpower through the coupledstructures, from the location of the excitation. An advantageof thepower flow technique isthat, if one componentof thestructure is modified, the analysis need only be repeal;ed for the modified structural compo- nent,and the reevaluation of thecoupling expressions. The use of thestructural power flow method is demon- stratedfor beamlike and platelike structures including an analysis of theinfluence of structuraland excitation parameterson the responseof thecoupled structures. Power flow methods can also be usedto analyzemore thanone type of wavemotion, which is useful in caseswhere incident wave power is scattered into other wave typesnear structural discontinuities. Additionally, experimental results for structuralpower flow are dis- cussed.[Work supportedby NASA.]
9:55
CO,.Coupled systems of anisotropiclayers: A groupformulation. Michael Schoenberg (Schlumberger-Doll Research,Old QuarryRoad, Ridgefield, CT 06877-4108)and FrancisMuir (Departmentof Geophysics, StanfordUniversity, Stanford, CA 94305-2215) A matrixformalism allows for thesimple calculation of theanisotropic, homogeneous medium equivalent to a stationarydistribution of thin layersin weldedcontact. Each layer itself may be an elasticanisotropic medium.The calculatedmedium is equivalentto the layeredmedium in the longwavelength limit. The physicalproperties of anyconstituent of theheterogeneous system can be shown to betransformable to an elementof a commutative group. Adding group elements Gn andG B (correspondingtothin layers of constitu- entsA andB, respectively) gives the group element for the homogeneous medium equivalent to the interleaved layersof A andB. The groupformulation, which is essentially a statement of conservationlaws applicable to layeredmedia, enables us to "decompose"an anisotropicmedium into several constituent sets of layersby successiveadditions of inverseelements, i.e., subtractions.If, after eachsubtraction, the remaininggroup elementcorresponds toa stableanisotropic medium, a validdecomposition isobtained. Within the group of all anisotropicconstituents there are nests ofsubgroups, each subgroup corresponding toa symmetrysystem more restrictivethan the most general triclinic anisotropy. Given the symmetry systems to whichthe constituents belong,the subgroup structure reveals, a priori,the symmetry of theequivalent medium.
Contributed Papers 10:20 A modalanalysis framework was developed toidentify the dominant C5. Dominant parametersof weakly coupledsystems. Takeru Igusa characteristicsof weaklycoupled systems using perturbation analysis and (Departmentof Civil Engineering,Northwestern University, Evanston, modesynthesis techniques. Modal parameters were obtained to quantify IL 60208) four types of characteristics•tuning,coupling, nonclassical damping,
$7 J. Acoust.Sec. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S7 and spatialinteraction. These parameters were subsequently used to de- M. Kerwin [J. Acoust.So<:. Am. $1, 952-962 (1959)], basedon thin- terminethe responseof thesystem to broadbandexcitation. It wasfound platetheory for thebase plate, a hybridmodel of a fluid-loadedplate was thatthe parameters could be used to isolatecertain modes of the system developedwhereby only the baseplate is describedby exactelasticity with dominanteffects on the response.The responsecharacteristics were theo• [P.S.Dubbelday, J. Acoust.SOc. Am. Suppl.1 80, S121(1986) ]. alsoexamined in thehigh-frequency range by extendingasymptotic mod- A descriptionbased on exact elasticity theory for all threelayers, without al analysis[E. H. Dowell and Y. Kubota,J. Appl. Mech. 52, 949-957 fluidloading, was also formulated [D. W. Fausett,L. V. Fausett,and P.S. (1985) ] to includethe asymptoticproperties of the coupledsystem pa- Dubbelday,J. Acoust.Soe. Am. Suppl.1 80, S121(1986) ]. The present rameters.The resultswere used to illustrateapplications in designof isola- studyfurther explores the interactionof viscoelasticdamping and fluid tion mountsin the low-frequencyrange, and in providinginsight into loadingby a descriptionaccording to exactelasticity theory of all three statisticalenergy analysis in the high-frequencyrange. The work is an layers,combined with fluidloading. For flexuralwaves, the modelshows extensionof thedynamic analysis of lumpedmass secondary systems sub- reasonableagreement with theearlier hybrid model at intermediatefre- jectedto low-frequencyexcitation [T. Igusaand A. Der Kiureghian,J. quencies,but discrepancies occur at lowerand higher frequencies. These Eng.Mech. 101,20-41 (1985)]. arediscussed, and possible mechanisms for thebehavior are brought for- ward. [ Work supportedby ONR. ]
10'38 11:05 C6. Effect of foundation design and resilient mount positions on C8. General relationshipsbetween acoustic impedanceand modal structure-bornenoise transmission.Peter K. Kasper and Russel density. Jerome E. Manning (Cambridge Collaborative, Inc., D. Miller (NKF Engineering,Inc., 12200Sunrise Valley Drive, Reston, Cambridge,MA 02138) VA 22091 } A generalformulation of theacoustic impedance is set out using mod- A typicalstructure-home noise control configuration consists of a al analysis.It is shownthat the averageacoustic resistance can be general- vibratingmachine, a setof resilientisolation mounts, a foundationsup- ly expressedin termsof themodal density of theacoustic space, the bulk portingstructure, and a buildingfloor or deck.If thefloor is represented complianceof thespace, and the average joint acceptance.Examples are by an infiniteflat plateand the foundationis comprisedofstructurai beam givenfor bothsimple and complex sources. Although the work presented elements,the acoustic power transmitted to the fioorcan be calculated in offersno previouslyunknown solutions to acousticradiation problems, it exactfashion using the dynamicdirect stiffness technique. This analysis providesa veryconvenient engineering result that canbe usedin statisti- approachinvolves the representation ofbeam segment properties in terms cal energyanalysis and other related acoustic analysis techniques. of dynamicfrequency-dependent stiffness coefficients. The solutionof dynamicdisplacement response to a harmonicinput forceparallels the solutionof staticdeflections to an appliedconstant load. Calculations of 11:20 transmittedacoustic power were made for a rangeof foundationstruc- tural parametersand resilient mounting configurations. The resultsindi- C9. An analyticmodel of sheacand moment interaction of reinforcingribs cate the sensitivityto suchdesign parameters on structure-bornenoise with cylindricalshells for acousticradiation. Courtney B. Burroughs transmission. (AppliedResearch Laboratory, Pennsylvania State University, P.O. BOx 30, StateCollege, PA 16804)and KennethJ. Becker(Department of the Navy, Naval SeaSystems Command, Washington, DC 20362-5101) 10:50 In responseto vibratorymotion of a cylindricalshell, reactive forces C7. Fluid-loadedcoastrained-layer analysis by exact elasticity theory. and momentsare generatedby reinforcingribs attachedto the interior surfaceof the shell.These reactive forces and momentsmodify the shell Laurene V. Fausett (Department of MathematicalSciences, Florida motion and therebythe farfieldacoustic radiation from the shell. Equa- Institute of Technology,150 W. University Boulevard,Melbourne, FL 32901} and Pieter S. Dubbeiday (Underwater Sound Reference tionsgoverning the motion of a rib, modeledas a circularring, are derived Detachment,Naval ResearchLaboratory, P.O. Box568337, Orlando, FL that includenormal and circumferentialshear forces, and momentsap- 32856-8337) pliedto theouter edge oft.he ring where the ring is attached to theinner surfaceof the shell.Approximate solutions for the impedancesof the ring The constrained-layerdamping technique is basedon the attachment at theouter edge are obtained and used in equationsof motionfor a point- of an absorbingviscoelastic layer and a stiff constraininglayer to the excited,fluid-loaded, circular cylindrical shell reinforced by periodically surfaceof a metalstructure to bedamped. Starting from the modelby F. spacedribs.
TUESDAY MORNING, 17 MAY 1988 WEST BALLROOM B, 8:30 TO 11:45 A.M.
SessionD. Underwater AcousticsI: StochasticVolume and Boundary Scattering I
Terry Ewart, Chairman /lppliedPhysics Laboratory H•V-10. Universityof Washington,1013 N.E. 40th Street,Seattle, Washington98105
Chairman's Introduction--8:30
Invited Papers
g.•
D1, Acousticvolume scatteringin the ocean--Approachingmaturity. B.J. Usciuski(Department of Applied Mathematicsand TheoreticalPhysics, Univc•ity of Cambridge,Cambridge CB3 9EW, England) Until recently,the approximationsof Born and Rytov werethe standardmethods used to describethe intensityfluctuations that arise when a soundwave is scatteredin thebody of theocean. Since these methods
J. Acoust.Soc. Am. Suppl. 1, Vol. 83, Spring 1988 115thMeeting: Acoustical Society of America arevalid only for verysmall fluctuations, it is notsurprising that agreement with experiment was often very bad.Over the last few years new analytical expressions forthe moments of the acoustic field, valid even for large intensityfluctuations, have been derived on thebasis of theparabolic wave equation. These results have been verysuccessful in explaining experiment observations and have also proved to be a powerfultool in furtherinc our understandingof fluctuationphenomena. From this pointof view,acoustic volume scattering can be regardedas a problemfor whichsome answers are now found. The expressionsare for thesecond moment, or complexcoherence, and the fourthmoment, which gives the intensityfluctuation spectrum. The sameforms havebeen derived by severalindependent methods and exist in the generalcase when a deterministicsound- speedprofile leads to curvedray paths.The analytical intensity spectrum has been tested by comparisonwith accuratenumerical solutions of the fourthmoment equation and the agreementis excellent even for verystrong scatter.The intensity fluctuation spectra observed in theexperiments at Cobb Seamount are well described by theirnew fourth moment expressions, where Rytov's approximation had proved hopelessly inadequate. It is stronglyrecommended that smallamplitude approximations, such as that of Rytov,should no longerbe used to describemultiple scatter observations when the newertheory, tested and valid in theseconditions, is now available.
9:00
D2. The numericalapproach to understandingrough surface scattering. Eric I. Thorsos(Applied Physics Laboratory,College of Oceanand FisherySciences, University of Washington,Scattic, WA 98195) Numericalmethods for solvingsimplified rough surface scattering problems "exactly" provide new in- sightsinto the roughsurface scattering problem. Exact solutions for I-D surfaceswith the pressurerelease (Dirichlet) boundarycondition are found with integralequation and Monte Carlo techniques. These studies providea betterunderstanding of the validity of the two basic approximations used in roughsurface scattering: theKirchhoffapproximation and first-order perturbation theory. From this work, a resolutioncan be given for the differencein the Kirchhoffand first-order perturbation theory predictions at smallsurface heights and at verysmall surface slopes, where both approximations have generally been considered valid. Studies with tbe Kirchhoffapproximation show that shadowing corrections based on geometricoptics lead to inaccuratepre- dictions.This work indicatesthe needfor tractablemultiple scattering theory for low grazingangle applica- tions.Comparisons with exact results show that "phase perturbation" theory predictions account for a signifi- cantpart of the multiple scattering associated with shadowing. ]n additionto better understanding theory with numericaltechniques, the accuracyof approximatenumerical methods can also be examined. Attempts arc nowbeing made to includerough surface scattering in marchingsolutions of the parabolicwave equation (PE). Methodsof verifyingthe accuracy of marchingsolutions using exact and PE integralequations will be brieflydiscussed.
9:25
D3. A tutorial reviewof the role of scatteringtheory in detectionand estimationof sonarsignals. Arthur B. Baggeroer(Departments of Oceanand ElectricalEngineering, Massachusetts Institute of Technology, Cambridge,MA 02139), T. Ewart (Applied PhysicsLaboratory, University of Washington,Seattle, WA 98105), andJohn McCoy (CatholicUniversity, Washington, DC 20064) Thereare many theories that describe time and Doppler spreads, coherences and intensities, and higher- orderprobability densities of scatteredsignals. The sonarsignal processor concerned with detectionand estimationapplications must use parts of thesetheories to designsystems and analyze their performance. This talk will givea tutorialreview of howscattering theories are used by thesonar signal processor.
9:50
1)4. The statisticsof scatteredwaves. E. Jakeman(Royal Signalsand Radar Establishment,Malvern, WoreshireWR 14 3PS, England) Thestatistical properties of wavesscattered by random media arc a subjectof long-standing interest both in thecontext of remotesensing and noise limtation of systemperformance. It is usually agreed that when many independentscattering centers contribute to theamplitude at thedetector, then the statistics will be Gaussian asa consequenceof the central limit theorem. There are many situations where this is not the case, however, andit isnecessary toadopt non-Gaussian noise methods. In thepast, these have often been chosen on the ad hoc basisof goodnessof fit to limiteddata sets, rather than lacing well founded on physicalinsight and a clear understandingof the scattering mechanisms involved. More recently, it hasbeen conjectured that hierarchical behaviorwithin the scattering medium, which is often observed in nature,could give risc to rathercharacteris- tic non-Gaussianfluctuations. Supporting evidence from optical scattering experiments, from microwave sea echodata, and from certain exactly solvable configurations has encouraged the development ofthe K-distribu- tionmodel. A generalreview of thisarea will be presented.
S9 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S9 ContributedPapers
10:IS ly developedrenormalized perturbation theory. Overall, this approxima- tion is satisfactorybut, sinceit is a singlescattering theory, it failsto DS. Commentsregarding the two-scaleexpansion for the fourthmoment incorporatemultiple scattering effects that are smallin its domainof va- of wavespropagating in randommedia. R. S. Patton and J. L. Codona lidity. (AT&T Bell Laboratories,Whippany, NJ 07981) Over recentyears, the methodof two-scaleexpansions has been ap- pliedwith a greatdeal of successto theproblem of thefourth moment of 11:00 wavespropagating in randommedia. Specifically, the methodyields a remarkablysimple expression for the two-pointintensity correlation that DS. Roughsurface scattering using the phase-perturbationtechnique. isvalid in bothstrong and weak scattering and both high and low spatial Shira L. Brosehat(Applied PhysicsLaboratory and Departmentof frequencies.Unfortunately, of the publishedderivations, there is no clear ElectricalEngineering, University of Washington,Seattle, WA 98195), presentationof the physicalcontent of the methodnor a straightforward Eric I. Thorsos(Applied PhysicsLaboratory, College of Oceanand approachfor findingeither corrections to theapproximation or general- FisherySciences, University of Washington,Seattle, WA 98195), and izationsfor treatingwaves in arbitraryrefracting media. The clearest ap- Akira Ishimaru(Department of ElectricalEngineering, University of proachwould appear to bebased on standard configuration-space path Washington,Seattle, WA 98195) integi'als[13. I. Uscinski,C. Macaskill, and M. Spivack,J. SoundVib. 106, Thereis a needfor a solutionto theproblem of wavescattering from 509-528(1986) ]; however,that derivation appears to have(at themini- roughsurfaces that is accurate when both the classical field perturbation mum) a differentphysical content from the other,moment equation andKirchhoff (physical optics) approximations are not. In thiswork, the based,derivations. The natureof the approximationis discussedin the validityof thephase perturbation technique is examinedfor a regionin contextof bothconfiguration-space andphase-space path integrals. The parameterspace when the two classicalsolutions fail. Numerical results two-scaleapproximation is foundto be a simpleinterpolating formula for thephase perturbation scattering strength are compared with exact betweena highspatial-frequency approximation (found by considering numericalresults for one-dimensionalpressure-release surfaces having a high-frequencyasymptotics inphase space) and the low spatial-frequency Gaussianroughness spectrum and Gaussian height and slope distribu- behaviorfrom the standard Born approximation. Looking at the approxi- tions.Exact results are obtainedusing a MonteCarlo technique. It is mationfrom this perspective lends both physical insight and simplifies the foundthat in theregion considered the phase perturbation results agree derivationto thepoint where it becomesquite simple to writeexpressions withthe exact results over all scatteredangles away from low grazing for a for refractingenvironments and correctionterms. fixedangle of incidence.Furthermore, in manycases it is foundthat ex- changingincident and scattered angles in thephase perturbation equa- tionsgives an alternatephase perturbation solution whose numerical re- 10:30 sultsare in excellentagreement with exact results. [Work supported by ONR. l D6. Non-Kirchhoffelastic wave scatteringfrom rough interfaces. HenrikSchmidt (Massachusetts Institute of Technology,Cambridge, MA 02139) and W. A. Kuperman(Naval ResearchLaboratory, 11:15 Washington,DC 20375) D9. Efficient,high-frequency, parabolic-equation modeling with surface An earlierdeveloped solution technique for seismoacoustic scattering scatteringloss. Martha E. M. Head (Naval OceanographicOffice, Bay by stochasticallyrough interfaces [W. A. Kupermanand H. Schmidt,J. St.Louis, NSTL, MS 39522-5001), EleanorS. Holmes,Lewis B. Dozier Acoust.See. Am. 79, 1767(1986) ] hasbeen modified to treatthe scatter- (ScienceApplications International Corporation, 1710 Goodridge Drive, ingwithout applying the Kirchoff approximation. This is accomplished McLean,VA 22102),and W. Jobst(Naval Oceanographic Office, Bay byincluding the rotation of theboundary conditions in the perturbation St. Louis,NSTL, MS 39522-5001) formulationand by numerically evaluating the scattering integrals, allow- ingthe roughness correlation lengths to befinite. It isdemonstrated that Solvingthe parabolic approximation tothe acoustic wave equation by theeffect of thescattering on themean (coherent) field can be accounted thesplit-step algorithm isone of the principal methods for estimating the forin a self-consistentmanner by solving the unperturbed problem with acoustic-pressurefield in a range-dependentunderwater environment. modifiedboundary conditions. As was the case for the Kirchhoff approxi- Thealgorithm iscomputationally intensive and is usually practical only at mation,the non-Kirchhoff formulation has been implemented in theseis- lowfrequencies. A modification to thesplit-step algorithm is provided moacousticSAFARI code [ H. Schmidtand F. B. Jensen,J. Acoust.Sec. (the calculation-frequencymethod), which allows rapid calculation of Am. 77, 813 (1985)], allowingfor simultaneoustreatment of multiple transmissionloss at highfrequency, includes surface and bottom loss, and roughinterfaces. The model is applied to analyzethe effect of scattering providesfor theredistribution of surface-scatteredenergy in wavenumber froma roughshallow water sea bed as well as from an Artic ice canopy. By space. comparingthese results to theKirchhoff results for the same problems, it is demonstratedthat the non-Kirchhofftheory predictssignificantly higherlosses, in particularfor acousticwavelengths of thesame order of magnitudeas the roughnesscorrelation length. [Work supportedby 11:30 NRL.] D10.Integral equations for therough surface scattering amplitude. John A. DeSante (Departmentof Mathematics,Colorado School of Mines, 10:4•, Golden. CO 80401 )
D7. Kirchhoffapproximation for scatteringfrom a roughfiuid-•olid The roughsurface scattering amplitude can be writtenas a surface interface. Dalcio K. Dacol (U.S. Naval ResearchLaboratory, Code integralover the boundary values of the fieldand its normalderivative on 5160,Washington, DC 20375-5000) the surface.A generalprocedure is describedto deriveLippmann- Schwingerintegral equations in k spaceon these scattering amplitudes for A generalprescription for the Kirchhoffapproximationfor wavescat- Dirichlet,Neumann, and interface or transmissionproblems. The proce- teringfrom a penetrablesurface is implementedfor the caseof acoustic dureis validfor arbitraryincident fields. It yieldsa multiplescattering scatteringfrom a roughfluid-elastic solid interface. A simplerversion descriptionof the scatteringprocess in terms of a fundamentalsurface developedfrom this prescription produces results that compare favorably interactionfunction that describesthe scatteringprimarily as a phase with exactnumerical results for the caseof simpleperiodic fluid-solid modulationdue to theroughness. Simple examples of singleand double interfaces.This approximation iswell suited to describesound scattering scatteringfor a one-dimensionalsurface will bepresented. Also included from a randomlyrough fluid-solid interface. In this case,the resultsob- arecomments on the diffractionand scattering problems that arisewhen tainedusing this approximation are compared with those from a previous- treatingunbound surfaces. [Work supportedby ARO.]
810 J. Acoust.Sec. Am. Suppl. 1, Vol. 83, Spring 1988 115thMeeting: Acoustical Society of America S10 TUESDAY MORNING, 17 MAY 1988 GRAND BALLROOM A, 9:00 TO 11:50A.M.
Session E. Architectural Acoustics I and Noise I: Interior Noise of Aircraft
RichardJ. Peppin,Cochairman Scantek,Inc., 51 MonroeStreet, Rockville, Maryland 20850
DanaS. Hougland, Cochairman David L. Adams Associates,Inc., 1701 Boulder Street, Denver, Colorado 80211
Chairman's Introduction--9:00
Invited Papers
9:05
El. The architecturalSlmce: A comparisonof acousticalcriteria usedto assessaircraft noisein roomsand aircraftinteriors. Dana S. Hungland(David L. AdamsAssociates, Inc., 1701Boulder Street, Denver, CO 80211 )
The aircraftas a soundsource is viewedin differentways depending upon the locationof the receiverof concern.Although, often, the samesource is dealtwith, thereare definite differences between the impacton a buildingoccupant near an airportand the crewor passengerswithin the actualaircraft. Studies of both receiversuse similar primary criteria such as speech intelligibility, fatigue, annoyance, and hearing loss. The criteriadiffer in relativeimportance, length of exposure, and perception of value. The application of secondary criteriaused in noise control studies to quantitatively express attenuation through the exterior shell, acceptable backgroundlevels, and room effect, vary between the two fields of study.The comparison of similaritiesand differencesinexisting, commonly used criteria for these two fields of studyis explored. Suggestions forareas of furtherstudy and correlationbetween the two fieldsof studyare investigated.
9:30
E2. Aircraft interior noiseresearch at the NASA LangleyResearch Center. Clemans A. Powelland Richard J. Silcox(NASA LangleyResearch Center, Mail Stop463, Hampton,VA 23665-5225) The StructuralAcoustics Branch at the NASA LangleyResearch Center has been a majorcontributor to thefield of aircraftinterior noise research for thepast 15 years. In additionto contributionsfrom its in-house program,the branch has supported many contributions from other organizations through an active grant and contractresearch program. Although the current emphasis ofthe program is on advanced turboprop airplanes andhelicopters, the goal has always been to develop and improve interior noise prediction methodology. This allowsfor theincorporation of appropriatecontrol measures in newaircraft at thedesign stage rather than throughthe use of add-onacoustical treatments with potentially large weight penalties and reduced effective- ness.The proposed presentation will concentrate on some of the more recent basic research activities concerned with the transmissionof noiseinto aircraftinteriors through both airborne and structurebornepaths. These includemodeling of thestructural response, the interior acoustic response, energy flow techniques, and acous- tical-structuralcoupling. In addition,some results of recent activities in activenoise control will bepresented. Althoughthe emphasis of thepresentation will beon researchconducted in-house, some research conducted undercontracts and grants that is closelyconnected to in-houseactivities will alsobe discussed.
9:55
F_3.Interior noise characteristics of rotorcraft.Charles R. Cox (Bell HelicopterTextron, MD-11, P.O. Box 482, Fort Worth, TX 76101 ) Themajor noise sources and weight-efficient means for noisecontrol in differentsize, single-rotor helicop- terswere determined by numerousmeasurements. Noise inside helicopters ischaracterized by numeroushigh- frequencytones during all flightmodes and by broadband wind noise at cruiseairspeeds. The tones, emanating primarilyfrom the main transmission, the engine gearbox, and accessories, are the most objectionable and are presentin the cabinof a helicopterat muchhigher levels than in the cabinsof fixed-wingaircraft. Because of thischaracteristic, certain criteria used to determineacceptability, particularly speech intelligibility, are found inappropriatewhen applied to helicopterinterior noise. Noise control is realized by design changes, e.g., source reductionand sound transmission path interruption, and by tailoredsoundproofing treatments. While some of thesedesign changes are generic, most are uniqueto a particularhelicopter model. This reflectsthe multiple
Sll J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America Sll structuraldesigns concepts, the variety of drive system mounting configurations, andextensive use of compos- itematerials. Emerging challenges involve the noise control of all composite rotorcraft airframes and of the new tilt rotor aircraft.
EA. Advancedacoustic test methodsused in aircraft interior aenust'cdesign. Mark A. Lang (McDonnell Dou81asCorporation, Douglas Aimraft Co.,•MC 36-80,Long Beach, CA 90846) Advancedmeasurement techniques, particularly sound intensity measurement, are becoming an important designand diagnostic tool in aircraftdesign. The utilizationof thesetools in aircraftmeasurement presents uniquechallenges that mustbe overcometo obtainmeaningful data. Once the dataare obtained,careful analysismust be utilized in orderto interpretthe measurement results. A workinggroup under SAE Commit- tee A-21 is currentlydrafting a documentintended to provideguidelines on soundintensity measurement underflight conditions. This document will bediscussed in thelight of somerecent sound intensity measure- menksmade in theDouglas Aircraft Fuselage Acoustic Research Facility as well as under flight conditions to highlightsome of themeasurement challenges involved in thistype of measurement.
10:4•
ES. Tes• procedurefor improveddata quality in aircraft interior noisemeasurements. Curtis I. Holmer (NoiseControl Technology, Inc., 2440Freetown Drive, Reston,VA 22091) Vehicleinterior noise measurements, and especially aircraft measurements, present a testenvironment that is inimicalto considereddata gathering due to the expenseof operatingtime and the fact that the vehicle characteristicschange during the measurement. A methodfor dealingwith thisis to constructa measurement procedurethat providesfor independentevaluation of dataquality after the measurement is complete. One suchmethod relies on the useof the relationshipbetween the integralof intensityon a closedsurface and the powerproduced or consumedwithin. Performing the measurement topermit estimation of this integral and an independentestimate of the powerprovides the opportunityto checkdata validity.This paperexplores the approachin detailand examines the resultingdata for a seriesof in-flightmeasurements. The measurements wereperformed on theGulfstream IV aircraftas part of a programto definenoise source areas. Value of the techniqueis exploredas a functionof severalparameters including amount of absorptionwithin the closed surface,reactivity of the measurementvolume, and other significant parameters. [Work supportedin partby GulfstreamAerospace Corporation, Savannah, GA.]
11:10
l•.6.1ntensitymeasurementsforinterioraniseenntrol.Robert g Cohen(BceingCommercialAirplanes, P.O. Box 3707, M/S 01-41, Seattle, WA 98124)
Typicalnoise sources, purposes of makingintensity measurements, and difficulties in makinguseful inten- sity measurementsin jet aircraft interiorsare surveyed.Primary noisesources contributing to cabininterior noisearc boundary layer, air conditioning,engines, and machinery. Transmission paths include both air and structure.Primary applicationsof intensitymeasurements in evaluatingand controllingthe effectsof these sourceson cabinnoise include ( 1) locatingand ranking sources for noisereduction; (2) optimizingacoustic treatmentfor minimizingweight; (3) providingdata in supportof cabinnoise predictions related to planned changesin structureor acoustictreatment; and (4) transmissionloss measurements for evaluatingand im- provingsidewall panel acoustic effectiveness. Measurement difficulties include limited flight time, require- menksfor samplingdetail in unknowngradients across each surface, unknown surface absorption, and un- known backgroundnoise levels interfering with the measurementof surfaceradiation, determining the accuracyof eachmeasurement, and interpretingthe data especiallyas a functionof data quality.
ContributedPaper
11:35 cabinnoise due to multiplecorrelated point forces, such as those acting on E7. Usingreciprocity to predictaircraft interiornoise due to multiple enginemounting points. The transfer functions are determined by a reci- correlatedinput forces. !stvan L. Ver (BBN LaboratoriesIncorporated, procalexperiment by measuring, atthe force excitation points, the magni- 10 Moulton Street, Cambridge, MA 0223g), William H. Mayes tudeand phase of thevibration response due to theoperation of a point soundsource located in thecabin. The experiments reported were carried (Aircraft Noise ReductionDivision, NASA Langley ResearchCenter, outon the fuselage of a lightaircraft. The methodology, the experimental Hampton,VA 23665), and MichaelC. McGary (BoeingCommercial Airplanes,Seattle, WA 98124) apparatus,the software,and the testresults will be presentedand the advantagesof the reciprocitymethod discussed. [Work supportedby This paperreports on how to utilizereciprocity to predictaircraft NASA LangleyResearch Center. ]
S12 J.Acoust. Soc. Am. Suppl. 1, Vol. 83, Spring 1988 115thMeeting: Acoustical Society ofAmerica S12 TUESDAY MORNING, 17 MAY 1988 METROPOLITAN BALLROOM, 9:00 A.M. TO 12:00 NOON
SessionF. Educationin AcousticsI: UndergraduateStudent Projects (PosterSession)
Les Atlas, Chairman Departmentof ElectricalEngineering, FT-10, Universityof Washington,Seattle, Washington 98195
ContributedPapers
All posterswill bedisplayed from 9:00 a.m. to 12:00noon. To allowcontributors the opportunity to seeother posters,contributors of odd-numberedpapers will beat theirposters from 9:00 to 10:30a.m. and contributors of even-numberedpapers will be at their postersfrom 10:30a.m. to 12:00noon.
Fl. Perceptualevidence for anticipatoryassimilation of adjacentstops. Thiswork studied the principal modes of vibration of a baritoneguitar $effrcyW. Chanand Amy Dolcourt-McElroy(University of California, withsteel strings tuned to A•, D2,G2, C3, E3, and A3, a fifthbelow those of Berkeley,1777 Euclid Avenue,Berkeley, CA 94709) theusual folk guitar. Likewise, the corresponding resonances are designed to bea fifth belowthose of a steel-stringfolk guitar [R. E. Rossand T. D. Prior workshowed that/akta/, createdby splicingtogether appropri- Rossing,J. Acoust.Soc. Am. 65, S72 (1979)]. The modalshapes have ateparts of/aka/and/ata/, arcperceived as/ata/when the stop closure beendetermined by scanningthe plates with an accelerometerand also by durationapproximates that of a singlemcdial stop; at longstop durations, scanningthe nearfield sound with a probemicrophone. Sound radiation bothstops are heard.Apparently listeners need to integratethe multiple fields in an anechoic room for several of the modes are also shown. acousticevents into a singlepercept that makessense for their language. At short closureduration, productionconstraints dictate that there is onlytime for onestop, hence the weakercues for the firststop are disre- garded.The present study reinforces this interpretation by showing that voicelessconsonant-spliced clusters, e.g.,/akta/, requirea longerclosure F4. An automatedportable transducer test system.Karl W. Rehn and intervalto be heardas two segmentsthan do voicedones, e.g.,/agda/, GregoryL. Cannon(Applied Research Laboratories, The University of becausein naturalspeech voiceless stop closures arc longerthan voiced Texasat Austin, P.O. Box 8029, Austin TX 78713-8029) stopclosures. Perception, then, is guidedby listeners'expectations that arc furtherbased on naturallyoccuring speech events. A setof computerprograms was developed to controlan automated portabletransducer test system. The system hardware, oonsisting ofcom- mercialtest equipment and a custom-designedamplifier/attenuator unit, wascontrolled by a Hewlett-Packardintegral personal computer. Pulsed F2. Effectsof varyingchord progressions on P 3 event-relatedpotentials measurements were used for all tests to decrease the effects of self-reso- in musiciansand nonmusicians.Sarah W. Chuang,Robert D. Frisina nancesand reflections,which can corruptcontinuous-wave (ew) mea- (Departmentof Physiology,University of Rochester,Rochester, NY surements.This computer-automatedtransducer test system offers the 14642),Gary C. Crummet (Departmentof Otolaryngology,University usera wide selectionof tests,such as immittaneeloops, impedance and of Rochester,Rochester, NY 14642), and JosephP. Walton (Eastman phaseplots, dc resistance(megohm) measurements, discrete frequency Schoolof Music, Universityof Rochester,Rochester, NY 14642) measurements,and Fl:rr analysis.Besides being able to performeach type of testsingly, the user can combine test types and setup parameters into Event-relatedpotentials were recorded from the scalps of bothmusi- "canned" routinesin which the systemexecutes a user-definedtest se- ciansal•d nonmusicians while actively engaging in an auditorydiscrimi- quence.The structure and content of theoverall test system are discussed, nationtask. Using a classical"oddball" paradigm [Donchin, Psyeho- anda qualitativediscussion of the sytem's efficiency, accuracy, and flexi- physiology18, 493 ( 1981) ], subjects(N = 19) listenedto oneseries of bility is given.The limitationsof the systemand sample test configura- puretones and four series of musicalchord progressions in which the tionsand output are also presented [Work supportedby NAVSEA.] secondof four chordsdiffered for the targetstimuli. The levelof discri- minabilityfor the progressionsrelied upon the degreeof changeof the targetchord in termsof melodicinterval, sonority, and voicing alteration in thealto and soprano lines. The P 3 waveformsdisplayed greater laten- ciesand widths for chordprogressions relative to thepure tones. For the F$. Nonlinearsound scattering of crossedfocused beams in the presence chordprogressions, musicians had a higherpercentage of hitscompared of turbulence.Stephen C. Rifeand Murray S. Korman(Department of to nonmusicians,with the formergroup showing a significantinverse Physics,United States Naval Academy,Annapolis, MD 21402) correlationbetween number of hits and P 3 latency.Additional observa- tionswere that the P 1potential for the first chord of theprogressions was Experimentsare performed involving the interaction of mutuallyper- considerablygreater in amplitudethan the P 1 for thethree later chords. pendicularcrossed ultrasonic beams overlapping and interacting in the Also,P 1 amplitudeswere greater for musicalsounds compared to pure presenceof turbulencein water.The turbulenceis created by a d = 0.64- tones,and the negative wave immediately following P 1 showeda larger cm-diamsubmerged water jet withnozzle exit velocity 13.3 m/s. A profile amplitudefor the puretones relative to the chords.The resultssuggest of sumfrequency pressure p + versusdistance, with characteristics of tur- that melodicinterval and voicingcues differentially affect neural poten- bulentvelocity, is detemined by scanningacross the width of thejet with tialsgenerated from central auditory and memory systems. Different cog- the twocw acousticbeams (of frequeneiesf•= 1.9MHz, f: = 2.1 MHz) nitivestrategies related to musicaltraining may underlie these observa- focusedon a pointin thejet. Thefocal lengths are 15.2 cm for both sending tions.[Work supported by Eastman School of Music, Program in Biology units.In scanning,the receiving transducer unit (unfocusedand located andMedicine, Audiology Division, University of Rochester.] 15.2cm from the interaction region) moves along with the senders so that propagationdistances never change. The receiveraxis is perpendicular to thejet. Scatteredsum frequency pressure profile scans (at distancesof 16d and34d fromthe nozzle) compare well with known values of theradial F3. Vibrational modes of a baritone guitar. Juan Lozano, Eric turbulentrms velocityprofile. Statistical properties of skewhesss and T. Watson,and ThomasD. Rossing(Northern Illinois University, kurtosisk fromfluctuations in p+ aremeasured as a functionof radial Dekalb, IL 60115) distance.[Work supportedby theNaval Academy Research Council.]
$13 J.Acoust. Soc. Am. Suppl. 1,Vol. 83, Spring 1988 115thMeeting: Acoustical Society ofAmerica S13 F6. Interaction of acoustic and thermally induced instabilities. A transducerreceiving unit (fixedrelative to thesender) is directed per- StaceyWest and CharlesThompson (Department of Electrical pendicularto thejet axis.This unit is unfocusedand is located15.2 cm Engineering,University of Lowell,One University Avenue, Lowell, IVIA from the focalpoint. The scatteringangle between sender and receiveris 01854) 45' in the "upstream"direction. Profiles of scatteredpressure p• as a functionof the radialscanning distance r arecompared to knownturbu- Thermal convection in a low Prandtl number fluid can exhibit both lentrms velocity profiles at distancesof 16dand34d fromthe nozzle exit. steadyand unsteady flow states. The stateto whichthe fluidis attractedis Measurementsof the spectralbroadening and Dopplershift of the scat- dependenton theratio of thebody forces to theviscous forces. This ratiois teredpressure taken at pointsacross the width of thejet will be usedto calledthe Rayleigh number. Above a criticalvalue of theRayleigh num- predictthe localvalues of the rms turbulentvelocity and the meanjet Derit hasbeen shown that unsteadythermal modes cut on and cut off in velocity,respectively, across the widthof thejet. [Work supportedby the discreteranges of temperature with increasing value of the Rayleigh num- Naval AcademyResearch Council.] ber. In thispaper, resuits ofan experimentalinvestigation into how high- frequencyacoustics waves interact with theseunsteady thermal modes are presented. F9. Acoustic remote sensingof the upper ocean. I. Shorey and N. P. Chotiros(Applied ResearchLaboratories, The Universityof Texas at Austin, P.O. Box 8029, Austin, TX 78713-8029) F7. An •lgoritha•for obf•iningimmitignee or impedam:eioo!•. Gregory There are particlessuspended in the ocean,such as bubbles and bio- L. Cannonand Karl W. Rehn (Applied ResearchLaboratories, The logicaldebris, which can backscatter acoustic waves. It is postulatedthat Universityof Texasat Austin, P.O. Box 8029, Austin, TX 78713-8029) the backscaRerfrom a collectionof particlesproduces a uniqueacoustic An immittanceloop algorithm has been developed for automatedso- signature.Since these particles are passiveriders, it wouldbe feasibleto nar transducertesting. This algorithmproduces an excellentloop given track the movementof the water by usinga sonarand a cross,correlation only two frequenciesseparated by a resonance.Measurements of imped- method.The volumescatterers and the operationof an acoustictracker anceand phaseare made at increasingfrequencies until the difference weresimulated on a computerin FORTRAN.Then a signalsimulator was betweenthe last recordedphase value and the currentphase value ap- usedto generatea set of backscatteredsignals for a multibeamsonar proachesa calculatedphase increment. At this point the new data are system.The scattererswere ensonified as their positions were changed in a recordedand a newphase increment is calculated.Measurements contin- way that simulatedflows and eddies in thewater. The signalsfrom succes- ue for increasingfrequencies until the specifiedupper frequencyis sivepings were cross correlated, and the resulting peaks in thecorrelation reached.When plotted as a loop,the recorded data can be used to deter- matriceswere used to estimate movement. A reasonablyaccurate portray- mineresonance frequencies, bandwidth, and other properties of resonant al of the water motion has been demonstrated. circuits.The algorithmhas beenimplemented on two computer-con- trolledmeasurement systems with excellentresults. Benefits of this algo- rithm include measurementof devicesthat have multiple resonances, measurementof high-Qdevices, wide freqeuncyrange of operation,very F10. Passive damping with differential density fluid damper. I. Bart, efficientuse of dataspace, and e•se of implementation.The algorithmand C. Morris, and M. Gormley (The Catholic University of America, itslimitations are discussed, and sample programs and output are present- MechanicalEngineering Department, Washington, DC 20064) ed. [Work supportedby NAVSEA.] This projectpresents a newclass of passivefluid dampers that usesthe differentialdensity between floating spheres and a dampingfluid to pro- videhigh damping characteristics over a widefrequency band. The devel- Fg. Scatteringof a focusedsound beam by turbulence.Mary M. lackson opeddamper is not only light in weightbut alsoeffective in damping and Murray S. Korman (Departmentof Physics,United StatesNaval multidi•octionalvibrations. These features make it superiorto the fully Academy,Annapolis, MD 21402) filledor the partiallyfilled fluid loopdampers that are •outinelyused to dampout thevibration of spinningspacecrafts and satellites. The effectof Experimentsare performed involving the scattering ofa cw beamof varyingthe designparmeters of the damperon its dynamiccharacteris. focusedultrasonic waves by turbulencein water.The turbulenceis creat- tiesis investigatedin an attemptto deternine the optimum values of these edby a d = 0.64-cm-diamsubmerged waterjet with a nozzleexit velocity parameters.The considereddesign parameters are, namely,the viscosity of 13.3m/s. A mechanicalapparatus positions a focusedbeam of sound of the dampingfluid and the concentrationof the floatingspheres. The (of frequency1.9 MHz} toscan across the width of theturbulent waterjet performanceof the damperalone is determinedwhen subjected to step plume.The sendingtransducer unit hasa focallength of 15.2cm andis andsinusoidal vibrations as well aswhen it is integratedin a largespace alwaysdirected at anangle of 450 to thejet axisand aimed "downstream." structure.
S14 J. Acoust.Soc. Am. Suppl. 1, VoL83, Spring1988 115thMeeting: Acoustical Society of America S14 TUESDAY MORNING, 17 MAY 1988 METROPOLITAN BALLROOM, 9:00A.M. TO 12:00NOON
SessionG. PsychologicalAcoustics I and Physiological Acoustics I: Experimental Techniques (PosterSession)
William M. Hartmann, Cochairman Departmentof Physicsand •lstronomy, Michigan State University, East Lansing, Michigan 48824-1116 Robert L. Smith, Cochairman Instituteof SensoryResearch, Syracuse University, Syracuse, New York13244-5290
ContributedPapers
All posterswill be displayed from 9:00 a.m. to 12:00noon. To allow contributors theopportunity tosee other posters,contributors of odd-numberedpapers will beat theirposters from 9.00 to 10:30a.m. and contributors oreyen-numberedpapers will beat theirposters from 10:30 a.m. to 12.-00noon. Contributors will beencour- agedto leavetheir posters in placeuntil 2:30 p.m.
GI. Someresearch "toys." Mead C. Killion (EtymoticResearch, 61 The third deviceis a phantomelectronic target simulator that digitizes Martin Lane,Elk GroveVillage, IL 60007) eachemitted click and then retransmitsthe click signalunder control of thecomputer. The developmentof thesethree devices has been a process An unbridledcuriosity and a certainlack of self-discipline have result- of evolutionin sophisticationover a periodof severalyears of dolphin ed in a seriesof productswhose research usefulness sometimes exceeds echolocationresearch conducted at the Naval OceanSystems Center. theirimmediate economic value to theauthor's company. On displaywill below-noise, low-vibration-sensitivity microphones for measuringspon- taneousand stimulated cochlear emissions; a l-in. microphone with swit- chablediffuse-field-inverse filter for simplifiedKEMAR' measure- G4. High-frequencyintensity-ealibrated threshold measurements using a ments;insert earphones with 70-100 dB of interauralisolation, 30-50 dB two-microphone system. William J. Murphy, Arnold Tubis of noiseexclusion, and variousfrequency responses (flat pressurere- (Departmentof Physics,Purdue University, West Lafayette, IN 47904), sponseat the eardrumto beyond12 kHz, flat diffuse-field-referencedre- and GlenisR. Long (Departmentof Audiologyand SpeechSciences, spouse,and TDH-39-1ike response ); and a !-mm-o.d.probe microphone PurdueUniversity, West Lafayette,IN 47907) with flat frequencyresponse beyond 10 kHz and two typesof precision tubing(soft silicone rubber with approximately35-dB wall attenuation As a meansof calibratinghigh-frequency hearing tests without the andsemirigid polyethylene with approximately50-dB wall attenuation }. problemsinherent in the useof pressure-calibratedheadphone systems [M. R. Stinson,J. Acoust.Soc. Am. Suppl.1 87, S75( 1987}], anintensi- ty-calibratedmethod of measuringhearing thresholds based on the two- microphonetechnique has been developed [W. J. Murphy, A. Tubis,and G2. Hand-heldauditory screener. Lynn S. Alvord (Departmentof G. R. Long, J. Acoust.Soc. Am. Suppl. I 87, S75 (1987)]. The two- CommunicationDisorders, University of Utah, 1201Behavioral Science microphoneprobe is usedto estimatethe acousticimpedance and the Building,Salt Lake City, UT 84112} powerreflection coefficient of the middleand inner ear system as well as to calibratepsychoacoustic measurements. Various cheeks on the reliabil- A newhand-held auditory screener is presentedthat utilizesfiltered ity of the long-termphase calibration are discussed. A computersimula- squarewaves as testsignals. Circuit designhas resulted in a total sizeof tion of the earcanal acoustics is usedto determinean appropriatedown- 2«X 4X 2«in., includingearphone and cushion. High-frequency filtering streamdistance of the probetips from the insert-earmoldsound port. of the squarewave signalresults in acceptablylow levelsof harmonic Typical data are givenfor both objectiveand psychoacousticmeasures distortionaccording to ANSI specificationsfor audiometers[ANSI S3.6- obtainedwith the two-microphonesystem. [Work supportedby a grant 1970].Low currentdrain design provides for a "battery-low"and "sig- from the DeafnessResearch Foundation. ] nal-on"indicator. Individual intensity pots allow for periodiccalibration or modificationof presentationlevels between 20 and 40 dB HL. Initial clinicaltrials with 50 patientsshow good agreement with conventional, pure-toneaudiometri 9 screening. Rationale for increaseduse of hearing GS. ModifiedDigi$ound-16 becomes a precisiondigital oscillator. W. screeningby healthcare practitioners is presented. M. Hartmann,D. L. Edmunds,T. V. Atkinson,(Physics Department, Michigan State University, East Lansing, MI 48824}, and Hal Chamberlin'• (MicrotechnologyUnlimited, 156 Wind Chime Court, Raleigh,NC 27615) G3, Instrumentationfor dolphinecholocation experiments. Whirlow W. L. Au (Naval OceanSystems Center, P.O. Box 997, Kailua, HI 96734) The Digisound-16is a two-channel,16-bit DAC-ADC systemwith a 32K word buffer.Because of its high performanceand low costit is an The useof personalcomputers has contributed significantly in dol- attractivesignal source for psycheacousticexperiments. The devicecan phinecholocation research by providingan inexpensiveinstrument to bemodified to serveas a precisiondigital oscillator by using the technique measureecholocation signals, monitor and control experimental devices, of fractionaladdressing [ W. M. Hartmann,J. Acoust.Sec. Am. 82, 1883- andto storedata. Dolphins typically emit short duration (50-[00/•s}, 1891 (1987}]. The frequencyresolution is 0.003 Hz over the audible broadbandclick signals with repetitionrates that canvary from tens of rangeand the total distortion approaches the theoreticallimit of - 85 dB clicksper second to several hundred clicks per second. In thispaper, three for single-channeloperation. The modificationrequires that the 15-bit electronicmeasurement devices used with Apple II computersin dolphin addressregister be replacedby a 24-bit increment-and-addcircuit. Two- echolocationexperiments will be discussed.The first devicemeasures the way communicationbetween the modifieddevice and the hostcomputer numberof clicksemitted, the intervalsbetween clicks, the peak-to-peak allowsthe address increment to bechanged continuously so that the oscil- amplitudeof each click, and the time of activationof variousswitches. latorcan be manually tuned, as in themethod of adjustment.The portion The seconddevice measures the frequencyspectrum (between 30 and 135 of thebuffer that is recycled can be reduced by factors of 2, underprogram kHz) of eachemitted click signal in realtime, with a resolutionof 15kHz. control,to as little as 2K words.The reducedbuffer length promotes
S15 J. Acoust.Soc. Am. Suppl. 1, VoL83, Spring1988 115thMeeting: Acoustical Society of America S15 efficiencywhile compromisingfrequency resolution and distortionfig- lr)Et,rrlFVis a programwritten in TurboPascal that accepts any com- ures.[Work supported by theNational Institutes of Health.] ') Present binationof up to 100speech waveform files as stimuliand up to 25 re- address:Kurzweil Music,411 WavcleyOaks Rd., Waltham, MA 02154. spousecategories. The waveform files can be from either ILS or C-Speech, or any compatibleformat. The programpresents the stimuliin block- randomizedorder and waits for the subjectto respondon thetouch sensi- G6. UsingMIDI, the standardMusical Instrument Digital Interface,for tive surfaceof a smalldigitizing tablet (Koala Pad). The tabletcan be psyehoaenustlcexperiments. H. E. F. Williams, G. L. Oibian, E. configuredto reflectany numberand patternof desiredresponses. The N. Hamden, and A. Evans (Physics Department,The American mostcommon case is whenthe numberof stimuliand responsesare the University,Washington DC 20016) same,producing a standardconfusion matrix. In this specialcase the programoutputs an additionaldata file that conformsto the formatfor MIDI, the musicindustry's standard Musical Instrument Digital In- terface,allows control of commerciallyavailable music synthesizers by theSpeech Information Transmission Analysis (SINFA) [Wang,Behar. Res.Methods lnstrum. g, 471-472 (1976) ]. A categorizationexperiment personalcomputers, as well as measurement and reproduction of musical can be run by specifyingmany stimulithat spana continuum,allowing performanceparameters. The resolution in frequency,amplitude, timbre, only two responses.The programrequires a PC or AT compatiblecom- and time of the systemis adequatefor variouspsyehoacoustie expad- puter,a Koala Pad connectedto a gameport, and a Data Translation meats.A YamahaDX7 synthesizercontrolled by an IBM PC wasused to DT2801-AD/A outputboard. [Work supportedby NIH.] measuresubjects' judgments of thesimilarity of synthesizedtones differ- ingin timbre[G. L. Gibianeta!., I. Aeoust.Soe. Am. Suppl.1 82, S68 ( 1987);G. L. Gibianet aL, AudioEngineering Society Preprint #2488 GII. Transducer and transducer measurementsystem for the PC. (1987) l- J.B. Allen (AcousticsResearch Department, AT&T Bell Laboratories, Room 2D530, Murray Hill, NJ 07974)
G7. DemonSration of auditory impedance/reflectancemeasurement The measurementof severalacoustic transducers using a measure- technique,and MIDI enters the research lab. Douglas H. Kecfe ment systemthat is implementedon the PC will be shown.This system (SystematicMusicology Program, School ofMusic, DN-10, Universityof usesthe ARIEL DSP-16 anda PC4300. The softwaredeveloped for this Washington,Seattle, WA 98195) purpose,which is called SYSid, was developed first about 8 yearsago, and A microcomputer-basedimplementation of a two-microphonewave- in the pastfew yearswas ported to the PC. SYSidwill measurethe fre- tube techniquefor measuringacoustical impedance and energy reflec- quencyresponse of a transducerin lessthan 100ms with an accuracythat tance will be demonstrated.Data can be fully analyzedwith graphical is determinedby eitherthe nonlinearitiesor the noisein the system.Typi- outputwhile the subjectis present,thus giving the researcherimportant cally, this noiseand distortionis more than 80 dB down, givingresults flexibility.The MusicalInstrument Digital Interface(MIDI) is a stan- thatare accurate to I partin 104.SYSid also will measure the phase, group dardin electronicmusic with potential research applications in theareas delay,impulse response, and the distortion as a functionof frequencyand of acousticalstimulus generation, instrumentation control, and data col- level.In thedemonstration, an attemptwill bemade to measurethe acous- lection.Selected applications will bedemonstrated on themicrocomputer tic impedanceof a sectionof tube,and a rho-cscreen. with MIDI peripheraldevices, and a MIDI softwarelibrary designed for integrationinto the researchsetting will be described.[Work partially supportedby NINCDS. ] GI2. Softwarefor analyzingand plottingscientific data. Wait Jesteadt, StephenT. Neely,Brian P. Callaghan, and Roger L. Ferrel(BoysTown National Institute, 555 N. 30th Street, Omaha, NE 68 i 31 )
Gg. A programmablesubject re•ponso box for psyehophysicul Our PUTprogram, in usein a numberof laboratories,translates ahigh- experimentation.Robert Ling and Edward M. Burns (Departmentof leveldata plottingcommand language into Tektronixor HP-compatible Speechand HearingSciences, University of WashingtonJG-15, Seatfie, output.It isspecifically designed for scientificapplications, allowing max- WA 98195) imum flexibilityin the useof partiallog axes,labels, and error bars,with rapidoutput on a graphicsterminal or publicationquality hardcopy on a Softwarethat utilizesa commerciallyavailable touch pad ("Koala pen plotter or laserprinter. aREa, a spreadsheetdata analysisprogram Pad") as an all-purposesubject response box will be demonstrated. developedin parallelwith PUT,allows data to berepresented in termsof a [Work supportedby NINCDS. ] five-dimensionalfactorial design. It providesconvenient data entry from either keyboardor digitizer pad, automatichandling of missingvalues, data transformations,low-level statistical analyses, and flexiblegenera- G9. A computerinterface for psychophysicaland speech research with tion of tablesand plots.Plots are generatedby producingeLT input files. the Nucleuscocldear implant. Robert ¾. Shannon,Doug D. Adams, GREGcan call usersupplied programs to fit specificmodels or perform RogerL. Ferrel,Robert L. Palumbo,and Michael Grandgenett(Boys other functions,so the data analysisframework can be easilyexpanded. Town National Institute, 555 N. 30th Street,Omaha, NE 68131) Both PUXand •EG run under RT-i 1, TSX,uNIx, and MS-nOS;they will be demonstratedon a portableAT clone.[Work supportedby NIH.] A custominterface was designed and implementedthat, underpro- gram control,allows presentation of any availablepulse stimulus to a patientfitted with theNuelens implant. The interfaceconnects to a stan- dardparallel output port of a PC or AT compatiblecomputer. The host GI3. Applicationof digital signalproeesalng techniques to rasearchon computersends a streamof bytesto the parallelport that specifythe soundlocalization.Frederic Wightmanand Doris Kistler (Department configurationof the desiredoutput pulses. Upon reedptof the data,the of Psychologyand WaismanCenter, University of Wisconsin,Madison, interfacegenerates the appropriatepulse-coded sequence to deliverthe WI 53705) ß,peeified pulses to thepatient's external coil. This interfacemakes it possi- The availabilityof fast, general-purposelaboratory computers has bleto interleavepulses on two or moreelectrodes, to modulatethe ampli- dramaticallychanged the way researchis donein manyareas of psycho- tude or timing of a pulsesequence, or to sweepa stimulusacross the acoustics.In studiesof soundlocalization, for example,standard digital electrodearray. This interfaceallows investigatorsto conductpsycho- signal-processingtechniques are usedto synthesizestimuli for headphone physicaland speech experiments that cannotbe performed with the stan- presentationthat contain all of the localizationcues (e.g., interaural dif- dard speechprocessor interface (SPI) normallyused to setthe pafient*s ferences,pinna effects) that are availablefrom soundspresented in free device.[Work supportedby NIH.] field.This bringsto localizationresearch a much neededdegree of stimu- lus control and specificity,and allows us to addresscertain basicissues that were heretoforeinaccessible. This presentationwill surveythe gen- GI0. IDENTIFY:A programfor speechrecognition, identification, and eral principlesinvolved in the applicationof digitalsignal processing to categorizationexperiments on PC compatiblecomputers. Robert localizationresearch, and will discussthe practical limitations of thespe- V. Shannon, Robert L Palumbo, and Michael Grandgenett (Boys cifictechniques that are used(e.g., FFrI'S,FIR filters,inverse filtering). Town National Institute, 555 N. 30th Street,Omaha, NE 68131) Empiricaldata will beshown in orderto addressquestions relating to the
S16 J. Acoust.SO•. Am. Suppl. 1, VoL83, Spring1988 115thMeeting: Acoustical Society of America S16 classictradeoffs among signal/noise ratio, bandwidth,word-length, and GI7. The use of multidimensional scaling techniques for revealing samplingfrequency. The feasibilityof real-timedigital processing in cur- perceptualcategories for complexstimuli in animals.Robert i. Dooling, rentlyavailable PC-based systems will alsobe discussed. [Work support- Kazuo Okanoya,Susan D. Brown,and ThomasJ. Park (Department ed by NIH, NSF, and NASA. ] of Psychology,University of Maryland, CollegePark, MD 20742)
A combinationof operantconditioning and multidimensionalscaling techniquesfor demonstratingnatural perceptualcategories for complex G14.Classical respiratory conditioning used in auditorypsychophysics of soundsin smallbirds arc described. Birds arc trained using operant condi- the goldfish:Methods and resultsin detectionand discriminationstudies. tioningprocedures on eithera same/differentdiscrimination task or on a R. Fay,S. Coorobs,and C. Whccles(Parrely Hearing Institute, Loyola task requiringthe detectionof changein a repeatingbackground. Re- Universityof Chicago,6525 N. SheridanRoad, Chicago, IL 60626) sponselatcncics arc usedto constructsimilarity matrices,and multidi- mcnsionalscaling procedures arc then usedto produccspatial maps of Classicalconditioning of respirationhas beenused to study many complexsounds reflecting perceptual organization. Stimulus similarity is aspectsof hearingin the goldfish,including absolute and masked thresh- representedby spatialproximity. Stimulus groupings in multidimensional olds, frequency,intensity, and time discriminationthresholds. A re- spaceindicate perceptual categories that canbe confirmed by clusteranal- strainedfish is presented with a 7-sauditory signal that ends with a brief yses.These procedures have been used to studythe perception of complex electricshock across the body.Shock causes an unconditionedsuppres- soundssuch as bird callsand speechin smallbirds, but thesetechniques sionof respirationlasting several seconds. Several pairings of theauditory shouldalso be usefulin examiningthe perceptionof complexsounds in signalwith the shock results in respiratorysuppression during the signal. other animals.[Work supportedby NIH.] Initial conditioningisrapid (conditioned responses appear within the first 10 to 20 trials), severalthresholds can be obtainedin oneday, and thresh- olds can be obtainedfor individual animals for severalyears. Critical factorsfor successwith this methodinclude the measurementof respira- tion (usinga thermistor),method of animal restraint,levels of electric GI8. Small.sample statistical analysis of Levitt's psychophysical shock,intertrial intervals, false alarm estimation, overall respiratory rate, procedure.Brent W. Edwardsand Gregory H. Wakefield(Department subjectselection, and water conditioning. Details of the methodsand pro- of E]IectricalEngineering and Computer Science, University of Michigan, eeduresthat havebeen found useful in conditioningand thresholddefini- Ann Arbor, MI 48109-1109) tion will bc givenalong with illustrativedata on thresholdvalues and Lcvitt's2IFC adaptiveprocedure is a widelyused psychophysical stabilityin maskingand intensitydiscrimination experiments. methodfor estimatingdetection or discriminationthreshold. The com- monlyaccepted analysis shows that the averageof the Icyelsat which reversalsoccur approaches the 0.707 point on the psychometric function whenthe levelsarc updatedaccording to the "2-up, l-down"rule. This G15. Psychoacoustical measurements with delays in neonates' analysisis based on a continuousrepresentation ofthe psychometric func- vocalizations. Lincoln Gray (Department of Otolaryngology, tionand on the asymptotic behavior of the tracks. The more practical case Universityof TexasMedical School, Houston, TX 77030) wasconsidered in whichthe psychomctric function is represented by a set of discretelevels and in whichthe stopping criterion is a fixednumber of Newbornchickens momentarily delay their usually incessant peeping reversals.A Markovmodel for thereversal levels is proposed as a method whenthey hear either an onsetor changein an auditorystimulus. This fordetermining the statistics of theestimator. Based on this approach, the unconditionedresponse provides several ways to studythe earlydevelop- estimatoris significantlybiased for smallnumbers of reversals,and de- ment of hearing. (1) Receiveroperating characteristics, drawn from creasesmonotonically as this number becomes large. Convergence to the pooledhistograms of responseson stimulusand controltrials, strongly 0.707point, however, is not guaranteed and depends on the sampling of suggestthat thisresponse is a measureof auditorydetection. Areas under thepsycbometric function. These results hold, in general, regardless ofthe thesecurves can be usedto quantifythe auditoryabilities of newborn formof thepsychometric function or themethod by which the psycho- subjects.(2) Adaptiveprocedures based on pairedcomparisons of stimu- metricfunction is sampled. More rapid convergence with less bias may be lusand control trials provide rapid estimates of thresholdsand difference achievedby tailoring the sampling to thepsychomctric function. Several limens.A five-frequencyaudiogram, for example,can bc obtained from a approachesto the design of suchsampling will be discussed. [Research newbornchick in severalminutes. (3) Responsesto all possiblepairs of supportedby AFOSR.] transitionsbetween multiple stimuli can be usedas proximitydata for multidimensionalscaling algorithms. This providesa "map"of howaudi- tory perceptionschange immediately after birth. (4) Neonares'responses to naturalisticstimuli may be differentthan those to morearbitrary pure- toneand noise signals. Delayed vocalizations also occur in otheranimals, GI9. Positronemission tomography (PET) as a techniquefor studying allowingcomparative studies ofauditory development. [ Work supported patternsof regionalcerebral function associated with auditory perception by NIH.] and speech production. John J. Sidtis, Vijay Dhawan, James R. Mocllcr, StephenC. Strothcr, David Eidelberg,and David A. Rottenberg(Department of Neurology,Memorial Sloan-Kettering Cancer Center, New York, NY 10021) G16. Techniquesfor free-fieldtesting of spatialattributes of acoustic signals.R. Wayne Gatehouse(Department of Psychology,University of PET isan in vivotechnique for the measurement ofpatterns of regional cerebralblood flow (rCBF), metabolicrate for glucose(rCMRGIu), and Guelph,Guelph, Ontario N 1G 2W 1, Canada) theuptake of ncurotransmittcrssuch as dopamine. Scanning times range For a numberof yearsGatehouse and his coauthors have reported on from I rainfor rCBF, to 2 h fordopamine, with rCMRGIu scansrequiring studiesof variousaspects of spatialacoustics (localization, depth percep- I h. Studiesof rCMRGIu in patientswith movementdisorders and tion, masking)done under free-field conditions of signalpresentations. dysarthriahave demonstrated correlations between voice onset time ab- Basically,the paradigmshave involved comparisons of the responsesof normalitiesand rCMRGIu in thebasal ganglia; similar studies with dopa- normallyhearing, artificially degraded (mortaural or binauralocclusion mineare currentlyin progress.Because of the shorterscan time, rCBF by car muffsand plugs),and real hearingimpaired subjects of various studiesarc better suited for perceptual studies. Using a steady-stateC0150 typesand degrees of loss,under conditions that presumablymirror the inhalationtechnique, 14 I-rain scanscan be obtained back-to-back. The morenormal hearing environment, i.e., reverberantor semireverberant C0150/PET hasbeen used to studyrCBF patternsin within-subjectstud- rooms,all typesof signals(noises, tones, speech), from positions that vary ies that includeboth stimulatedand unstimulatedperiods, with stimuli in both azimuth and elevation and from all around the subjects. In this such as broadband white noise, music, and complex tonesin a pitch dis- session,some of thesetechniques and conditionswill be displayedin a criminationtask. Results to datedemonstrate that bothregional and glo- moredetailed manner than is usuallyavailable in verbalpresentations of balchanges are associated with stimulationstates. PET providesa unique themethodologies. [Work has been supported under various NSERC and techniquefor studyinghuman cerebral physiology associated with audi- MRC grants.] toryperception and speech production. [Work supportedby NIH. ]
S17 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S17 G20. Acetyicholineeffects on toneelicited single unit responsesin rat etersfor eachfilter were decided by takinginto account observed physio- auditorycortex. H. K. Rucker (Departmentof Physiology,Meharry logicaldata and filter stability.The parameteroptimizing effects of the MedicalCollege, Nashville, TN 37208) filter were discussed.Sound spectrograms produced by usingthe filter- bankshowed that this auditory filterbank model allows excellent frequen- Urethane(1.5 g/kg) anesthetizedmale Sprague-Dawley albino rats cy and time resolution.A hardwareimplementation of the filterbank, wereaffixed with a headbolt and a burr holeplaced over the temporal composedof floating-pointDSP chipsand a hostcontroller, was studied. cortexand woundedges infiltrated with lidocainebefore being placed in In additionto the computationalfiltering functions, some control func- an electricallyshielded, sound attenuating chamber. Tone burstswere tions,such as automatic addressing previously done in thehost controller, generatedwith a softwarecontrolled M308 stimulusgenerator and 300 are distributedto eachDSP. Therefore,control of eachDSP unit by the stimuluscontrol module (Modular Instruments,Inc. ) and sentvia a sec- hostis easier. More sophisticatedfunctions are easily added. This is in part ondaryamplifier to a Realistic40-1377 loudspeaker. Sound-pressure lev- due to a programmemory commonly accessible both from eachDSP as elswere determined with a calibrated•-in. B & K 4135condensor micro- well as the host.Finally, the auditoryfilterbank is capableof a frequency phone.Single unit activitywas recorded via the centerbarrel (4M NaCI) analysisof up to 3000 channelswithin 5 ms, which is equalto the time of glassmulti-pipette. Drug barrels containedacetylcholine chloride requiredfor thewave to travelthe basilar membrane. It isalso expandable (0.SM, pH 4.5), afropinesulfate (0.02M, pH 5.0), and glutamicacid to allowhardware implementation with morethan 30 000 cascadedDSP (monosodiumsalt, 0.2M, pH 8.0). Sixty-sixpercent (n = 21) of units units. characterizeddemonstrated an attopine sensitive Ach facilitated response to nearbest frequency tone bursts. An attopineinsensitive Ach induced responsedecline was noted in 13% (n = 4). [Work supportedby NSF Grant BNS 8617937.]
G23. Whistling with heavy gases or at elevated pressures.R. G21.A newset of auditorydemonstrations. A. J. M. Houtsma(Institute Stuart Mackay (San FranciscoState University,San Francisco,CA for PerceptionResearch, P.O. Box 513, 5600 MB, Eindhoven,The 94132) Netherlands),T. D. Rossirig(Department of Physics,Northern Illinois University,DeKalb, IL 60115),and W. M. Wagenaars(Institute for Mostdivers can speak easily but cannotwhistle at elevatedpressure. PerceptionResearch, P.O. Box 513, 5600 MB, Eindhoven,The Suggestively,sperm whales are not knownto whistle.However, rigid Netherlands) flutesor bird callsremain easy to blow"at depth";frequency changes little.The authorcan just whistle tunes at 4 arm (equivalentwater depth A new seriesof auditorydemonstrations for classroomuse, modeled of 30 m). At oneatmosphere he could whistle with a breathof halfair and afterthe we!l-known "Harvard Tapes" [ D. M. Green,Harvard Laborato- half sulfurhexafluoride but not with pure SF6 (densityrelative to air of ry of Psychophysics(1978) ] wasdeveloped and produced. The demon- 5.1); alternatewhistling in andout aidsthe observation.Possibly this is strationsare issued on a compactdisc (CD), areaccompanied by a 92- due to increasinggas density yielding damping vibrations in the soft tis- pageexplanatory booklet, and fit convenientlyin a handyslide box. suesshedding vortices. Caution: Limiting duration prevents suffocation; Almostall demonstrationswere synthesized digitally on DEC Vax 11/ possibleimpurities in SF6 are toxicand corrode metal. The SF6lowers 780and MicroVax II computers,converted to analogsignals with DSC- whisperand speechtones but not hum and whistlefrequencies; reflex 200 16-bit two-channelD/A converters,and recordedon 16-bit PCM changesin the throatcan confuse the latter.A colleague,Jori Pegg, can videotape. Brief spoken introductions to eachdemonstration were re- whistletunes normally as well as through his nose with mouthclosed and cordedon similarPCM videotape. Editing of themaster tape was done vocalcords constricted to a slit.His nose whistle quenches at 20 m andthe digitallyin 1630format on t-in. U-Matic tape, from which the CD master normalone at 10 m, suggestingsites of differentstiffness. It is saidto be wasmade. The CD medium,allowing a S/N ratioof upto 90 dB,is very impossiblefor organpipes to soundif the streamis denserthan the sur- appropriatefor sounddemonstrations that require a cleanacoustic back- roundings.Several blew well (frequencylowered) with octafluoropro- ground.The mediumalso provides random access to individualdemon- pane (density6.5) in air. strationsor partsthereof, allows preprogramming of any desired combi- nationof demonstrations,and is muchmore resistant against wear and tearthan traditional tapes or re•ords.The CD canbe played on any CD playercapable of handlingselections from up to 80tracks. The demo set is availablethrough the ASA.
G24. Digital synthesisof binauralauditory localization •imuth cues using headphones.Richard L. McKinley and Mark A. Ericson G22. An auditoryfilterbank design and its hardwareimplementation with (Armstrong AerospaceMedical Research Laboratory, Wright- DSP. Takashi Koreakineand TatsuyaHirahara (ATR Auditory and Patterson AFB, OH 45433) VisualPerception Research Laboratories, Twin 21 Building,MID Tower 2-1-61Shiromi, Higashi-ku, Osaka 540, Japan) A laboratorydemonstration prototype of a digitalauditory localiza- tion cuesynthesizer has been developed. This synthesizer uses a single As the first stepin constructinga signalprocessor to simulatethe audioinput that is separately processed in real-time for independentpre- humanauditory function, a computationalfilterbank with cochlearfre- sentationto eachear using headphones. The headphone presented acous- quencyanalysis function was designed IT. Komakineand T. Hirahara, tic signalsare easy to localizeand appear to beout of head.The acoustic Tech. Rep. IEICE $pn. SP87-45,65-72 (1987)(in Japanese)]and its imageis stabilized for headmovement by use ofa threc-spacehead track- hardware with DSP implementationcompleted. From the variouscoch- ing device.The paperwill describethe salientparameters of the design.A !earfiltering models proposed to date,a cascade/paralleltype filterbank descriptionof the psychoacousticand electroacoustic measurements that model[e.g., R. F. Lyon,ICASSP82, 1282-1285 (1982) ] waschosen fol- ledto the design will bepresented. Human performance data on free field, lowedby designof a filterbankcomposed of 61-channelIIR filterblocks. simulated,and synthesized localization cues will bedescribed and a real- Eachfilter block was spaced at «Bark intervals from 1 to 21 Bark. Param- time interactive demonstration will be available for interested listeners.
S18 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S18 TUESDAY MORNING, 17 MAY 1988 EAST BALLROOM B, 9:15 TO 11:45 A.M.
SessionH. EngineeringAcoustics I: Transducersand Their Applications
GeorgeS. K. Wong, Cochairman PhysicsDivision, National ResearchCouncil, Montreal Road, Ottawa, OntarioK1A OR6, Canada
Caroline Fu, Cochairman BoeingAerospace, P.O. Box 3999, MS:82-58,Seattle, Washington98124
ContributedPapers
9:15 A methodfor determiningthe free-field open-circuit voltage sensitiv- ity of a reversibleunderwater electroacoustic transducer from the differ- H1. Fiber optic flexural disk microphone.T. Hofler, D. A. Brown, and encein its inputelectrical impedance when loaded by waterand air was S. L. Garrett (Physics Department, Naval PostgraduateSchool, investigatedtheoretically and experimentally. An equationfor the sensi- Monterey,CA 93943) tivity wasderived using reciprocal two-port network theory. The theory An interferometricfiber optic microphone consisting of a 10-m-long, takes into account the diffraction due to the finite size of the transducer, 4-cm-diam,flat-wound (spiral) singlemode optical fiber bondedto a its finitemechanical impedance, and its free-fieldradiation impedance in simplysupported, 8-cm-diam, 3-mm-thick aluminum disk will be de- water.An experimentto testthe predictionsof the theorywas performed scribed.In the presenceof a pressuredifference across the plateor an usinga 6-in.-diam,hollow, piezoelectric ceramic spherical transducer. accelerationof theplate, the surface strain induced in theplate is transmit- Theresults of the experiment agreed within several dB with the results of a ted to the opticalfiber. The opticalphase shift inducedby the strain is standardcomparison calibration over the frequency range for whichboth detected in an all-fiber Michelson interferometer. The calculated strain the electricalimpedance and comparisoncalibration data are considered [ A. E. H. Love,A Treatiseon theMathematical Theory of Elasticity(Do- reliable. The calibration method described,which has been termed the ver, New York, 1944),4th ed.,Sec. #309] andoptical strain measured Delta-Zmethod, may be useful for in situmonitoring of transducersensi- usingstatic pressure, acoustic pressure, and a calibratedaecelerometer, tivityin installationswhich can be flooded and purged. a• Major, Canadi- arein goodagreement ( _+10%) andyield a sensitivityof 30 milliradians an Armed Forces.Permanent address: 415 Squadron,C. F. B. Green- perPascal per meter of opticalfiber below the plate resonance frequency wood, Nova Scotia BOP IN0, Canada. ß of 3 kHz. [Work supportedby theOffice of NavalResearch and Office of Naval Technology.]
10:00
H4. A four-sided flextensional transducer. John L. Butler (Image 9:30 Acoustics,Inc., P.O. Box6, North Marshfield,MA 02059) and Kenneth D. Rolt (RaytheonCompany, Submarine Signal Division, 1847West H2. Anomalous degrndation of electromechanicalcoupling and Main Road, Portsmouth,RI 02871-1087) resonancefrequency in tonpilz transducers.Michael P. Johnson (Acoustics and Sensor Systems Group, Systems Engineering A novel four-sided flextensional transducer has been constructed and Department,Gould, Inc., Ocean Systems Division, 18901 Euclid Avenue, testedfor low-frequencyoperation. The transduceris in theshape of an Cleveland, OH 44117) asteriodformed by fourcurved concave metal plates driven at theirjunc- tionsby four piezoelectricceramic stacks configured as the spokesof a In recentyears, the bandwidthof sonararrays has beenincreasing, wheel.As the stacksexpand in the positivecycle of operation,the four requiringwide bandwidthtransducer element designs. Tonpilz trans- curvedplates of theasterJori move outward in a motionthat is the sum of ducershave come to bethe dominantelement design in sucharrays. The the radialmotion and the outward bending of thecurved plates yielding a locationof thefronting blocks flexural resonance relative to theoperating cumulativeacoustic output. The experimentalmodel tested is approxi- band becomesa critical featureof thesewideband element designs. In mately25 in. (0.635 m) in diameterand 6 in. (0.152 m) highwith four somecases, unexpectedly low electromechanicalcoupling and resonance curvedsteel plates, 0.25 in. (0.00635m) thick,and operates in the fre- frequencieshave occurred. These results are typically attributed to "cross quencyband from 500-1500Hz. The resultsfrom the measurements modes"due to ceramicstack geometry or interfacecompliance at the show that current theoretical models for the conventional Class IV flex- stackends, though no conclusiveevidence exists for either.In a seriesof tensionaltransducer may be usedto estimatethe performanceof this modelinginvestigations utilizing the piezoelectric capabilities of thecom- transducer. mercial finite elementcode ANSYS, modeshapes at the fundamental resonanceof a numberof designswere compared. In certaindesigns, ex- cessiveand unsuspected bending motion in the frontingblock was found to degradethe electromechanicalcoupling and resonantfrequency. The resultsof theseinvestigations will bepresented along with somepossible 10:15 solutions. HS. A magnetohydrodynamicallydriven Helmholtz resonantprojector. C. E. D. Haney, J. T. Newmaster, and S. L. Garrett (Physics Department,Naval Postgraduate School, Monterey, CA 93943)
9:45 Unlike most conventionaltramducers that are noncompliant, a mag- netohydrodynamic(MHD) transduceris ultracompliantsince the volu- H3. Reciprocitycalibration of an underwatertransducer by the Delta-Z metricvelocity it generatesis entirely determined by theacoustical imped- method.R. Bedard'• and S. R. Baker (Departmentof Physics,Code anceof the loadit is driving [G. W. Swiftand S. L. Garrett, "Resonance 61Ba,Naval PostgraduateSchool, Monterey, CA 93943) reciprocitycalibration of an ultracomplianttransducer," J. Acoust.Soc.
S19 J. Acoust.Soc. Am. Suppl. 1, Vol. 83, Spring1988 115th Meeting: AcousticalSociety of America 819 Am. gl, 1619(1987) ]. To exploitthis property, a magnetohydrodynami- piezoelectrichydrophone within the resonatorvolume and by directob- tally drivenHelmholtz resonator has been designed and calibratedthat servationof themotion of thefluid in theneck using an MTI 1000-Fotonic consistsof a 400-mI-PVCtube with a 6-era-longLucite neck of rectangu- (fiberoptic) sensor.Operating at a resonancefrequency of about 100Hz lar crosssection that is placed in the 1.0-cmgap of a 0.2S-Teslapermanent with Q = 16,the transducerhas a sourcestrength of 5 cc/Amp-s. [Work magnetand filledwith salt water.The transducerwas calibrated using a supportedby the Officeof Naval Research.]
10:30-10:45
Break
10:45 11:15
H6. Transfer functions of piezoelectric transducersfor underwater Hg. Acoustic interaction effects in unbaffied transducer arrays. applicationsand their inverse problems.Shuchu Ren (Institute of Christian Audoly (Groupcd'Etudes ct de Rcchcrchcde DetectionSous- Acoustics,Academia Sinlea, Beijing, People's Republic of China) Marine Le Brusc, 83140 Six-Fours, France) Threerelated problems of thetransfer functions of piezoelectric trans- To calculatethe transmitting characteristics ofa transducerarray, it is ducersare solved:( 1) the theoreticalcalculation of the transferfunction, often necessaryto take into accountthe acousticinteraction effects. For (2) thereconstruction of thephase spectrum of thetransfer function from an arrayof pistonliketransducers set on planarsurfaces, the model most the amplitudespectrum, and (3) the extrapolationof the transferfunc- commonlyused is to assumethe pistons set in aninfinite rigid plane. This tionto the full frequencyrange. The transferfunctions of cylindricaland is not adequatewhen the array is not baffled,and when its dimensionsare sphericalpiezoelectric transducers are obtained by solving the scattering comparableto thewavelength. A theoreticalmodel, using the combined problemof the transduceras a soundreceiver. The calculationsshow that Hclmoltzintegral equation method, is developed to calculatethe mutual thesetransducers are typical minimum phase systems; consequently, the impedancematrix, the transmittingcharacteristics, and farfield directivi- phasespectra of their transferfunctions can be reconstructedfrom the ticsfor arbitrary antenna geometries. Some results are compared to analy- amplitudespectra using the Hilbert transform. For an arbitrary piezoelec- ticalmodels and to experiments.It is shownthat the unbaffiedcondition tric transducerit is impossibleto determinewhether it is a minimum producesimportant effects on the directivity functions and that the acous- phasesystem. To solvethis problem,an entirelydifferent technique to tic interactionsmodify the pressuredistribution on the surfaceof the reconstructthe phase spectrum isalso developed. The phase spectrum of array. the transferfunction of a F42-cspherical piezoelectric transducer is re- constructedfrom the measuredvalues of its amplitudespectrum. The resultsare consistent with the measured values of its amplitude spectrum. The extrapolationof the transferfunction of a piezoelectrictransducer fromthe finite frequency range to thefull range is done using the represen- tationtheorem of H pfunctions. For a sphericalpiezoelectric transducer, theresults obtained by extrapolationare in goodagreement with thetheo- 11:30 retical ones. H9. Compositestructure vibration analysisusing the finite element method.B. Hamonic,I. C. Debus,H. Toumeur (Institut Sup•rieur d'Electroniquedu Nord, 41 boulevardVanban, 59046 Lille Cedex, 11:00 France), and D. Boucher (Groupe d'Etude et de Recherche de D•tectionSous-Ma•ine, Le Brusc,83140 Six Fours les Plages, France) H7. Characterizationof length expandermagnetostrictive rare earth- iron rods under normal nse conditionsin transducers.F. Claeyssen, The needof low-frequencysonar transducers led to the conceptof D. Boucher,and S. Faure (GERDSM, DCAN Toulon, DCN, Le Brusc, flextensionalstructure and, currently, promotes the useof compositema- 83 140Six-Fours-Les-Plages, France) terials.To designthese transducers, a finite elementmodeling is very useful,because this method can accurately handle the variousconstitut- Magnetostrictiverare earth-ironrods are presentlymainly usedas ing parts as well as the radiationdamping. The finite elementcode activeelements of emissiontransducers. In this case,rods are mechanical- ATILA [J. N. Decarpignyet al., J. Acoust.Soc. Am. 78, 1499(1985)] ly prestressedand magnetically polarized. They are excited at highlevel hasbeen used to analyzevarious types of metallic shell flextensional trans- aroundthe transducer resonance frequency, which is much lower than the ducersand, morerecently, a specialthree-dimensional composite finite rod naturalfrequency. In orderto characterizethe rodsunder these con- clementhas been developed to take compositeshells into account.This ditions,a two-roddevice has been designed, on which the electricalim- elementis composedof unidirectiona!lyreinforced lamina and various pedancehas been measured before being processed by a computerpro- fiber orientations and materials can be used in the same element. The gram. The resultsare the longitudinalpiezomagnetic constants of the stiffnessfinite element matrix is calculatedusing equivalent elastic con- alloy.The programis basedon a plane-wavemodel; its assumptionson stantsIS. W. Tsai,"Composite Design," in ThinkComposites (Dayton, eddycurrents and leakage flux arediscussed with respectto resultspro- Ohio, 1985)] anda separatenumerical integration in threedirections for videdby a finiteelement computation. Coupling coefficient k33 , Young's eachlamina. This paperpresents the elementformulation and its tests. modulus,permeability, and lossterms have been reckoned using this Comparisonbetween numerical and experimentalresults for simple methodfor alloysof variousorigins and are presented.The characteriza- structureshas allowed the determinationof the physicalmodel param- tion of materialsstarted with L. N•cl's laboratoryalloy [1. Acoust.$oc. eters.Then, the modelingof a flextensionalshell has demonstrated the Am. $uppl. 181, S89 (1987) ] andis carried on with grain-orientedalloys. accuracyof this approach.
S20 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spdng1988 115thMeeting: Acoustical Society of America $20 TUESDAY AFTERNOON, 17 MAY 1988 EAST BALLROOM A, 1:00 TO 4:35 P.M.
SessionI: Bioresponseto Vibration I and Noise II: CombinedEffects of Noise and Vibration on Humans
Olavi Manninen, Chairman Departmentof PublicHealth, MedicalFaculty, University of Tampere,P.O. Box 607, SF-33101Tampere, Finland
Chairman's Introduetionml:00
In vited Papers
1:05
II.Complexenvironmentalexposuresandhearingfuuetions.Olavi Manninen (DepartmentofPublic Health, MedicalFaculty, University of Tampere,P.O. Box 607, SF-33101Tampere, Finland) This studydeals with changesin TT.q2 values,cardiovascular functions, haemodynamic activity, upright body sway,ratios of urinary catecholamines,and correlationsbetween these changes in complexexposure situations.The studywas based on a factoffalexperimental design with a totalof 12exposure combinations. Eachindividual experiment took 6 h with a pauseof I h at noon.The subjects(n = 60) wereexposed to noise andwhole-body vibration at two differentdry bulbtemperatures. The changeswere dependent on the combina- tionsof noise,vibration, and temperatureto whichthe subjectswere exposed. The TT$ 2 valuesat 4 kHz were associatedwith the haemodynamicindex (HDI) valueswhen the subjectswere exposed simultaneously to noiseand stochastic vibration at 35 øC.The TTS2values at 6 kHz wereassociated most strongly with the HDI valuesafter exposure to a combinationof noise,stochastic, or sinusoidalvibration, and a temperatureof 20 øC. The TTS2 valuesat 4 and 6 kHz correlatedpositively with the noradrenaline/adrenaline(NA/A) ratio when subjectswere exposed to noiseat 35 øC.The associationbetween the TTS• valuesand the 10A/NA ratio and especiallythe A/NA ratio wasvery strongwhen subjects had beensimultaneously exposed to noiseand sinusoidalor stochasticvibration at 35 øC.Furthermore, the highestpositive correlation coefficients were foundbetween the TTS2 values at 4 kHz andthe upright body sway values in theX directionwhen the subjects had beenexposed to noiseand sinusoidalvibration at 20 øC.
1:35
I2. Theeffeetsofnoiseandvibrationonacomplextask.J. SandoverandC.S. Porter (Departmentof Human Sciences,University of Technology,Loughborough LEI 1 3TU, England) Investigationsof the interactiveeffects of noiseand whole-body vibration on taskperformance have been doggedby suchproblems as the directeffects of vibrationon controland the resistanceof simpletasks to the environmentalstress. The paperdescribes an attemptto overcomesome of theseproblems and at thesame time to considerconditions of practicalrelevance. Subjects were exposed to heat,noise, and vibration singly and in combination.The environmentalmagnitudes were typical of someoccupations and exposureduration was approximately6 h. Subjectswere asked to performa visualvigilance and decisionmaking task, a visual monitoringand decodingtask, and a batteryof simpletasks. The formertasks were designed to placea significantcognitive load on thesubject and strategy was emphasized by performanceof bothtasks at the same time.The taskshad relevance to a realwork situation that thesubjects were used to. Subjectiveand physiologi- calmeasurements were also taken. The paperwill presentthe results of theseries of experimentsjust complet- ed.
2:05
I3. The influenceof whole-bodyvibration on noise-inducedhearing loss: A review of animal exl•eriments. Roger P. HamerniL William A. Ahroon, Robert L Davis (Auditory ResearchLaboratory, SUNY at Plattsburgh,Plattsburgh, NY 12901), and Donald Henderson(Communicative Disorders and Sciences, $UNY at Buffalo, Buffalo, NY 14260)
Thereis the suggestionin the literaturethat vibrationmay potentlate the effectsof noiseand may pose an increasedrisk of hearingloss. However, in humanexperimental studies, which, by necessity,are limited to low levelsof TTS, theeffects measured are consistent but relatively small. A verylimited number of animalstudies havealso shown an enhancedheaffng loss, but the scopeof thesestudies is limited by a large intersubject variabilityand a smallnumber of subjects.Also, the highlevels of stimulationthat wereused in someof these animalexperiments were not realistic.Our recentanimal studies (chinchilla) haveused a 30 Hz, 3-g-rmscage
S21 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115thMeeting: Acoustical Society of America S21 vibrationin combinationwith continuousnoise (95-dB, 0.5-kHz octaveband) and impactnoise ( 113-, 119-, or 125-dBpeak SPL) exposureparadigms. All exposureslasted for 5 days.The impactnoise exposures were designedto havean equaltotal energy.Temporary (compound) and permanentthreshold shifts were mea- suredusing evoked potentials. Sensory cell populations were evaluated with thesurface preparation technique. The resultsobtained from eachof the aboveparadigms were consistent in showingthat the presenceof vibra- tiondid nothave a statisticallysignificant effect on hearingthresholds. A parallelset of experimentsusing a 20 Hz; 2-g-rmsvibration is in progress.Preliminary conclusions are essentially the sameas thoseof the 30-Hz experiments.The suitabilityof the chinchillaas an animalmodel for usein vibrationexperiments will alsobe discussed.[ Work supportedby NIOSH.]
2:35
14, Combined effect of whole-body vibration and noise on the dopamine turnover in the rat brain, Akira Okada,Hiroyuki Nakamura,Hideki Nakamura,and Seiichi Nohara (Departmentof PublicHealth, Schoolof Medicine,Kanazawa University, Takaramachi 13- l, Kanazawa920, Japan) In orderto clarifythe combinedeffects of whole-bodyvibration and noise on dopamine(DA) metabolism withinthe brain,which is knownto regulatethe responseof the organismto variousstimuli, the DA turnover ratein regionsof the rat brainwas determined. The ratswere divided into fivegroups: ( 1) control;(2) whole- bodyvibration (4 G, 20 Hz, 90 rain) exposurealone; ( 3 ) noise[ 70 dB(A), 90 min] exposurealone; (4) noise [ 100dB (A), 90 min] exposurealone; and (51 combinedexposure for 90 minto whole-bodyvibration (4 G, 20 Hz) andnoise [ 100dB(A) ]. Changesof plasmacorticosterone levels were examined as indices of the pitu- itary-adrenalfunction (PAF). The whole-bodyvibration exposure alone caused increases in theDA turnover rate [an increaseof homovanillicacid (HVA) or HVA/DA] in the frontal cortexand nucleusaccumbens. Noise exposurealone caused metabolic increases in the amygdala.The combinedeffect of whole-bodyvibra- tionand noiseon the DA neuronsystems suggested that the responseof the PAF to thecombined stimulus was greaterthan that to eachstimulus alone.
ContributedPapers
3:05 calculatedby StevensMk. VII perceivedloudness method. The curvesare extendeddownward in frequencyto includethe two octavebands with I5. Resultsof periodicmedical examinations of workerswho are exposed meanfrequencies at 16 and 31.5 Hz. Finally, the high-soundlevel, low- to combined noise and vibration. Klaus Ruppc (Institute of frequencyregion of the curves[Blazier, Noise Control Eng. 16, 64-73 OccupationalMedicine, Humboldt-UniversityBerlin, Postbox 140, (1981) ] are alemarkedto showwhere human annoyance will probably Berlin 1040, GDR) and Gottfried Endcrlein (Central Institute of resultfrom vibrationscaused by suchnoise levels in contemporarybuild- OccupationalMedicine of the GDR, Berlin,GDR) ingconstruction. The paperdetails the use of theBNC curvesin writinga Synergisticeffects of combinedexposures to noiseand vibrationon specificationfor building construction and in determinationof theconfor- hearingloss have been recognized under laboratory conditions by some manceof the measuredresult to thespecification, or in ratingan existing authors.However, there arc few studiesof the chroniceffects of combined noise.Particularly important are the proceduresgiven for determining workplacefactors upon workers. An evaluationof the resultsof periodic "spectrumimbalance." The handlingof two typicaltypes of imbalance medicalexaminations of worker• who have beenexposed to combined arediscussed: ( 1 ) an acceptablespeech interference level accompanied by exposuresduring a great part of their workinglife showeda significant highlow-frequency band levels; (2) acceptablelow-frequencY band levels influenceof thecombination of factorsupon hearing loss. In a population accompaniedby a verylow speechinterference level. of 270 000 maleworkers, 52.6% wereexposed to noisewith levelsmore than 85 riB(A), 15.0% to whole-bodyvibration (WBV), and 4.1% to hand-arm vibration (HAV). Exposure to noise occurred in 70% of workersalso exposed to WBV, andin 80% exposedto HAV. Hearingloss (morethan 30 dB in 4 kHz) occurredin 16.7% of the groupof workers 3:35 with exposureto noisealone, but in 18.5% of the groupof workerswith combinedexposure to noiseand HAV. The prevalenceof medicalfindings 17.Vibrotactile intensity difference thresholds measured by two metbads. and a restrictedcapacity to work weresignificantly higher in the groups GeorgeA. Gescheider(Department of Psychology,Hamilton College, with combinedexposures. Clinton, NY 13323), Stanley $. Bolanowski,Jr. (Department of Physiology,University of KochesterMedical School, Rochester,NY 14642), and Ronald T. Verrillo (Institute for Sensory Research, SyracnscUniversity, Syracuse, NY 13233) The differencethreshold for the detectionof changesin vibrationam- 3:20 p]itudewas measured as a functionof theintensity and frequency of stim- I6. Building noise criterion curves, BNC, for interior spaces. Leo ulidelivered through a 2.9-cm2contactor to thethenat enfinence. Stimuli were either 25- or 250-Hz sinusolds or narrow-band noise centered at 250 L. Beranek (BBN Laboratories,l0 Moulton Street, Cambridge,MA 02238) Hz or widebandnoise. Thresholds were measured by two-intervalforced- choicetracking under two methodsof stimuluspresentation. In the two- Thispaper presents an updated set of noisecriterion (NC) curvesthat burstmethod, subjects had tojudge which of two 700-msbursts of vibra- are renamedbuilding noisecriterion (BNC) curves.They are based,in tion separatedby 1000ms wasmore intense. In the increment-detection part, on the latestdefinition of four-bandspeech interference level and on method,subjects had to detectan incrementin the amplitudeof vibration. "spectrumbalance," that is to say,the premisethat the loudnessesof all Thresholdswere consistently lower for detectingincrements in the ampli- bandscontaining the samenumber of critical bandsshould be equal, as tude of continuousvibration than in detectingamplitude differences be-
$22 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America $22 tweensuccessive bursts. Amplitude increment detection, however, was ncouslywith videofiuoroscopicdata taken while normal subjectsswal- relativelypoor when the standardstimulus was brief rather than contin- lowedsmall amounts of liquidbarium suspension and barium paste. The uous. The near miss to Weber's law was found for both sinusoidal and progressof the barium"bolus" could thus be followedradiographically, noisestimuli under both methods of stimuluspresentation. The difference and physicalevents in swallowingrelated in time to accelerometersignal thresholdwas not affectedby stimulusfrequency. [Work supportedby characteristics.The most prominentsignal feature is a relativelybrief NIH. I (200-ms)broadband noise that correspondsto the rapidpassage of the bolusthrough the lowerpharynx and cdevpharyngealsphincter into the esophagus.The spectrumof the noisecontains stronger high-frequency componentsfor a liquidthan for a pasteswallow. In closetemporal prox- imity to this noisecomponent, or evenmixed with it, is oftena periodic 3:50 signalthat is in thefrequency range of high-pitchedpronation (approxi- mately500 Hz) andwhich may be of laryngealorigin. Other low-ampli- 18. Determination of natural frequency of bone: Study of bone tudesignal features corresponded to structuralmovement of the hyoid/ abnormalities. Sanjay Yadav and V.R. Singh (Department of larynxor epiglottis.[Work supportedby NIH.] Instrumentation and Biomedical Ultrasonics, National Physical Laboratory, New Delhi-110012, India) Directand indirect methods have been used in thepast [ L. Nokes,W. J. Czyz. Mintow, I. Mackie, J. A. Fairclough,and J. Williams,J. Bio. Med. Eng.6, 4548 (1984) ] for thedetermination of naturalfrequency of 4:20 bone.Most of themare cumbersomeand costly. A simpleand quick ex- perimentaltechnique based on the stresswave propagation through bone I10. Changes in breath sound spectra with early developmentof the is developed.An acousticvibrator asa sourceof stresswave is usedto send human respiratory tract. Elzbieta B. Slawinski (Department of the stresswave through bone specimen under test, in thiscase, an in vitro Psychology,The University of Calgary, Calgary, Alberta T2N 1N4, sample,and a microphonepickup is usedto receivethe acousticsignal at Canada),D.D. McMillan (DepartmentofPediatries, The Universityof the otherend of the bone,which, in turn, givesan electricaloutput to be Calgary, Calgary, Alberta T2N IN4, Canada), and D. G. Jamieson measuredon a calibratedelectrical detecting system. The naturalfrequen- (Department of CommunicativeDisorders, University of Western cy of boneis determinedfor an optimumoutput with respectto different Ontario, London, Ontario, Canada) stresswave frequencies.A comparativestudy of normaland fractured Breathsounds in prematurelyborn infants are thoughtto be less"ve- boneis madeto diagnosethe sizeof the fractureand its rate of healing. sicular"than in terminfants. This differencein soundquality in newborns of variousgestational ages can be explainedby differenttransmission characteristicsof lungparenchyma. Chest sounds of 27 infants,of gesta- tional agesbetween 30 and 40 weeks,were recordedat differentchest locations.The resultsshow that thereis a strongcorrelation between chest soundsof normalbabies and the locationof the recordingmicrophone on 4:05 theinfant's chest. The spectraof inspirationof normalterm babies record- 19. Sounds of swallowing. Sandra Hamlet, Richard Nelson, and ed at the levelof the lowerposterior chest, show dominating amplitudes in Robin Patterson (Department of Otolaryngology,Wayne State the lowerfrequencies (350 Hz). Soundsrecorded at the upperposterior University,Detroit, MI 48201) chest,show spectra with enhanced amplitudes in a broaderrange of high- er frequencies(300-900 Hz). The studiesalso suggest the existenceof a Soundsof swallowingas detectedby a throat microphonehave been markeddifference between breathing patterns of prematurelyand term usedin the past primarily to mark the occurrenceof swallowing.The born infants.Sounds of prematurebabies tend to havea loweroverall sourceof thesesounds and what informationthe signalsmight contain intensity,and spectrawith dominantamplitudes in a narrowerband- aboutfunction is relativelyunknown. For thisinvestigation, signals from width. Moreover,dominant amplitudes are shiftedtoward higher fre- a miniatureaccelerometer taped to the throat were recordedsimulta- quenciesas compared to soundsrecorded in term infants.
TUESDAY AFTERNOON, 17 MAY 1988 EAST BALLROOM B, 1:00 TO 5:20 P.M.
SessionJ. EngineeringAcoustics II and Structural Acousticsand Vibration II: Structure--FluidInteraction Problems
Sung-HwanKo, Chairman Code2133, Naval .UndenvaterSystems Center, New London, Connecticut06320
Chairman's Introduction•l:00
Invited Papers
1:05
J1. Wave motionsin some compositeand layered media. Miguel C. Jungcr (Cambridge Acoustical Associates,Inc., 54 CambridgeParkDrive, Cambridge,MA 02 ]40) Fluid- andstructure-borne waves arc governed by the balancebetween elastic and inertiaforces embodied, respectively,in the complianceand density of the wave-bearingmedium. This paperis concernedwith a classof
S23 J. Acoust.Soc. Am. Suppl. 1. VoL83, Spring1988 115thMeeting: Acoustical Society of America $23 compositeand layeredmedia where compliance is dominatedby onemedium and densityby another.Exam- plesof this situation are: flexible conduits containing a liquid, e.g., blood vessels; bubble swarms; porous elastic solidswith Poisson'sratios in excessof one-third;compliant layers in contactwith a liquid,e.g., sedimentary oceanbottom containing gas bubbles or a cornpliantlycoated structure immersed in a liquid.
1:35
J2. The transientresponse of fluidqoadedstructures using time-domain methods. Peter R. Stepanishen ( Departmentof OceanEngineering, University of RhodeIsland, Kingston, RI 02881-0814) Overthe pastseveral years, time-domain methods have been developed to addressthe transientradiation and scatteringfrom structuresin fluidswith andwithout a meanflow. These methods are basedon usingin vacuomodal expansions with time-dependentcoefficients that are coupleddue to the fluid. To date these methodshave been used to investigatethe transientresponse of baffledmembranes, plates, and cylindrical shells.The developmentof the basiccomputational method will be reviewedand comparedto alternative methods,e.g., the doublyasymptotic approximation. Numerical results will thenbe presentedto illustrate typicaltransient phenomena of interest.A particularexample of interestis thedestabilizing effect of a mean flow overa vibratingsurface. [Work supportedby the Officeof Naval Research.]
2:05
J3. Measurementof flow excitationfunction. Charles H. Sherman,Sung H. Ko, and Barry G. Buehler (Naval UnderwaterSystems Center, New London,CT 06320) Wavevector filter measurements have been made with the objective of improvingquantitative knowledge of the flow excitationfunction. Two hydrophonearrays with the samenumber of elementsbut different spacingswere used. The moreclosely spaced array covers high wavenumbers characteristic of theconvective regionof the turbulent boundary layer pressure fluctuations, while the other array covers the lower wavenum- berregion where contaminants associated with the experimental arrangement also usually exist. The arrays weremounted on a dampedcomposite plate and tested at variousflow speeds. Individual hydrophone outputs fromboth arrays were sampled simultaneously, digitized, recorded, and spectral anaiyzcd in frequencyand twowave vector components. Results will be presented as amplitude contour plots as a functionof thetwo wave vectorcomponents on theplane of theplate at variousfrequencies. Results from the closely spaced array show theshape and level of the convective ridge and give an estimate of the difference between the convective peak leveland the low wavenumberdomain. Results from the otherarray show that the low wavenumberdomain alsocontains wave vector components associated with free plate vibrations {confirmed by calculations based onwork by D. J.Mead and S. Markus [J. Sound Vib. 10 (2) (1969)] } aswell as acoustic components. [Work supportedby ONR. ]
2:35-2:50
Break
2:50 J4.Plates with re•ular stiffedinginacousti½ media:Vibratiouandrediation. DenysL Mead (Department of Aeronautics& Astronautics, University of Southampton,Southampton, England) Flatplates and cylindrical shells with identical stiffeners at regularintervals constitute spatially periodic structures,and specially convenient methods of analysis are available for thestudy of theirvibrations. Some of themethods are suitable for theinclusion of theeffects of fluidloading from adjacent acoustic media. This paperoutlines the nature of thefree wave motion that can occur in periodicstructures that are stiflened either in onedirection or in twoorthogonal directions. It is shown how their responses todistributed sound fields can bedetermined by usingdisplacement functions consisting ofa seriesof space harmonics orof simple assumed polynomialmodes. The sound that is reradiatedor transmittedby the structure is alsofound. Methods that havebeen developed for analyzingthe response and radiation due to lineor pointforces are reviewed. Recent developmentsin theanalysis of periodicallystiflened cylindrical shells are described. The hierarchicalfinite elementmethod has been applied to determineflexural wave speeds in bothfiat reinforcedplates and in reinforcedcylinders. Symbolic computing has been used to setup therelevant stiffness and mass matrices. Somecomputed results are presented.
3:20 J$. Controlof soundradiation from submergedplates. Leonard Meirovitch and SurotThangjitham (Departmentof EngineeringScience and Mechanics, Virginia Polytechnic Institute & StateUniversity, Blacksburg,VA 24061) Thesubject of soundradiation from submerged elastic bodies, and in particularfiat plates, is our current researchinterest in the field of underwater acoustics. Because the plate is in contact with the surrounding fluid,
$24 J. Acoust.Sec. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S24 it givesrise to a generalclass of coupledproblems involving "radiation loading." This radiation loading modifiesthe motion of thesubmerged plate via the excess acoustic pressure, which, in turn,depends on the velocityof the plate. As a result,an interaction between the fluid and the plate exists in theform of "feedback coupling"[ M. C.Junget and D. Feit,Sound, Structures and Their Interaction (MIT, Cambridge,MA, 1986), 2rided.]. In manyapplications, theradiation pressure isan undesirable effect, so that the object is to suppress it. Thispaper is concerned with radiation pressure suppression through active control. The general idea of active controlof a structureis to suppressthe vibration by applying suitable feedback control forces [L. Meirovitch, Dynamicsand Control ofStructures (Wiley, New York, 1989 } ], whichrequires sensors capable ofmeasuring thestate of thesystem and actuators capable of applyingcontrol forces. A controltechnique capable of suppressingthe farfield radiation pressure isdeveloped.
ContributedPapers
3:50 to simplercases available elsewhere [i.e., IEEE J. Ocean.Eng. OE-12 (specialissue on Scattering),380-394, 395-403, and 419-422 (1987) ]. J6. V•brational response of water-loaded flat plate to a turbulent boundary layer pressurefield. Nilabh Narayan and Robert Plunkett ( Departmentof AerospaceEngineering and Mechanics,University of Minnesota,110 Union StreetSE, Minneapolis,MN 55455) The vibrationalresponse era 1.2- X40- X 200-mmflush-mounted fiat 4:20 plate, with heavyaccelerometers attached to the backside,to a turbulent boundarylayer (TBL) pressurefield wasmeasured at severalflow veloc- J8. Noise radiation from transverse air flow over cylinders of constant itiesin a 190-X 190-ramlow noise water tunnel. The measured response andvarying diameter: Experimental investigation of parameters.Karen spectraof thewater-loaded stainless steel fiat plate was compared with the C. Herdman and Paul R. Donavan (Noise and Vibration Laboratory, calculatedresponse in the400- to 1500•Hzfrequency range using the plate GeneralMotors Proving Ground, Milford, MI 48042} modalproperties and the pressurespectra. The phaseindicated convec- Aeoliantones from nonrigid, circular cylind•ers fixed on one end were tion velocityvaried with separationand frequency and was accounted for measuredusing sound intensity techniques. For cylinder diameters of the in the plate responsecalculations. The plate modalproperties were ob- orderof 3 to 4 mm andlengths of 750ram, sound intensity was measured tainedexperimentally by in situmodal analysis and the plate was found to 18 in. from the cylinderaxis at a directionperpendicular to the air flow. havea largenumber of heavilydamped modes in the samefrequency The radiatedsound intensity and Strouhal numbers were determined for range.The turbulentboundary layer pressure field driving the platewas severalcylinder diameters and flow velocities for the Reynoldsnumber measuredupstream and downstream from the plate and was found to be rangeof 4000to 11000. The sound intensity increased at a rategreater unaffectedby the vibratingplate. The two-pointpressure cross spectra at thantheoretical velocity to the sixthpower relationship IN. Curie,Proc. the wall for the TBL were found to vary with the axial and transverse R. Soc. London Ser. A 231, 505-514 (1955) ]. The influenceof resonant separationsin a way similarto that of the Coreossimilarity model and vibrationon the radiatedsound is examinedand comparedto previous followedthe multiplication hypothesis but hada differentform of explicit results[ Leehey and Hanson, J. SoundVib. 13, 465-483 ( 1971} ]. Signifi- dependenceupon the separation. cant reductionin Aeoliantones with minimalaxial taperwas demonstrat- ed.
4.•5 4:35
J7. Resonances in the sonar cross sections of shells near the sea surface, J9. Noise radiation from transverseair flow over cylinders of constant G.C. Gaunaurd (Naval SurfaceWarfare Center, White Oak Lab (R43 }, and varyingdiameter: An empiricalpredictive model. Paul R. Dooavan Silver Spring, MD 20903-5000] and M. F. Werby (Naval Ocean and Karen C. Herdman (Noise and Vibration Laboratory, General Researchand DevelopmentActivity, Code 221, NSTL Station, MS Motors ProvingGround, Milford, MI 48042) 39529-5004) A simplifiedmodel to estimate the amplitude and frequency content of A modelfor thestudy of acousticscattering from a spheroidalelastic the noiseradiated by constantand varying(tapered } diametercylinders shelllocated near a pressure-releaseinterface has been developed. The hasbeen developed. Using the measuredintensity level of Aeoliantones modelrests on themethod of imagesand on theextended boundary condi- for severalconstant diameter cylinders, the incrementalsource strength tion {EBC} approach.The methodyields a Tmatrix that mapsthe inci- along the cylinderhas been determinedas a functionof Reynolds dent field onto the resultingscattered field. The approachis usedto de- numbers.After choosingan elementsize consistent with the correlation scribe,compute, and display the echoesbackscattered from the shell,as lengthof thevortex shedding, tapered cylinders are modeled as a seriesof functionsof frequency[i.e., the sonarcross section (SCS) of the target] small,discrete cylinders each of constantdiameter. The total soundpro- andof angle[ i.e.,the differential scattering cross section (DSCS) ], andit ducedby the taperedcylinder is thendetermined by summingthe contri- permitsthe determinationof scatteringpatterns in the aporopriateres• bution of eachsegment over the c•dinderlength, properly accountin8 for naneefrequency regime. A goalof the workis to assessthe influenceera geometryusing the assumptionthat the sourcesadd incoherently.The nearbypressure-release boundary on the locationof the resonancefea- resultspredicted by this approachhave been found to be quantitatively turesin the SCSof submergedelastic shells. The modelpredictions are consistentwith the data measuredfor taperedcylinders at severalorienta- illustratedwith a varietyof pertinentnumerical displays exhibiting the tions. The model can also be readily extendedto examinethe effectof resonancechanges with increasing shell depths. The resultsare compared nonuniform flow on radiated noise.
$25 J. Acoust.Sec. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S25 4:50 5:05
J10. Applicationsof reactiveintensity. $. Adin Mann,III andJiri Tichy Jll. Water-structure interaction effects in stick-slip bearing vibrations (GraduateProgram in Acoustics,The PennsylvaniaState University, of a submergedframe structure.Hal Amick (BBN Laboratories,Inc., P.O. Box 30, StateCollege, PA 16804) 21120 VanowenStreet, Canoga Park, CA 91303) Low-frequencyvibrations of a hinged,submerged steel-frame struc- The reactiveintensity vector is a very usefulanalysis tool for the turewere observed during rotation at severalspeeds. The vibrations had acousticnearfield. Vibration sources,active and passiveabsorbers, and notbeen observed during operation in air priorto submersion.The vibra- diffractioncan be differentiatedby measuringthe reactiveintensity. The tionswere attributed to stickslip of the submergedhinge bearing caused realand imaginary parts of thespecific acoustic impedance should also be byhigh friction. The vibrations were measured and the characteristics of considered. Nearfield measurements will be discussed where the reactive thevibrations appeared to changewith rotational speed. A finiteelement intensityvector is used to determinethe location and distinguish between analysisof thegate, which included the "added mass" effects of thewater, pistonsources and a resonatorthat are acoustically coupled. The reactive disclosedthat the slip frequencies corresponded to several natural modes intensitymust be used because the time-averaged intensity vector does not of the gateframework. A finiteelement analysis of the dry structure distinguishbetween a resonatorabsorbing power and an activesource (without added mass) indicatedmuch higher frequenciesfor these absorbingpower. Diffraction will alsobe identified.These results will modes,suggesting that displacements associated with stick slip in thedry furthershow that all quantitiesmeasured with theacoustic intensity tech- structurewould not havebeen large enough to haveproduced observable niqueshould be analyzed. vibrations.
TUESDAY AFTERNOON, 17 MAY 1988 GRAND BALLROOM C, 1:00 TO 5:12 P.M.
SessionK. SpeechCommunication II: Cross-LanguageStudies: Acoustics, Phonetics, and Perception
JamesE. Flege,Chairman Biocommunications,University of Alabama,University Station, Birmingham, Alabama 35205
Chairman's Introduction--l:00
ContributedPapers
1:02 Previousstudies of nasality have concentratedon productional, acoustic,and perceptual features distinguishing nasal segments. Research KI. Voice onsettime and Fo perturbationunder conditionsof mixed reportedhere examines a wideraspect of nasality,namely its interaction airstreammechanism. fan R. A. MacKay and DariuszA. Zwierzynski with speakingfundamental frequency. To investigatewhether nasality (PhoentiesLaboratory, Linguistics Department, Universityof Ottawa, lowersor raisesSFF, bilingualspeakers recorded four passagesthat dif- Ottawa, Ontario K IN 6N5, Canada) fered (a) asto language--Frenchversus English--and (b) asto propor- Pulmonicand velaric airstream mechanisms are employed together in tionsof nasalsegments. Speakers' SFF wasmeasured in the four condi- the productionof clicks and accompanyingvowels. Questions thus tions.Statistical analysis of the resultsindicates that thereis significant emergeas to the temporaland microprosodiccoordination of thesetwo interactionof nasalitywith SFF, with sexof speakerand nativelanguage mechanisms,particularly with respectto thephonatory characteristics of alsobeing important variables. Implications of theseresults include: first, clicks.This reportis on the acousticmeasurement of the post-alvcolar segmentalinteraction with prosodicelements; second, in speechsynthe- clickin Sesotho,with particularreference to VOT andF o perturbations. sis,parameters for nasalsegments and F0 mayneed to beinterdependent, Thereare two phonemicallydistinct varieties of clickin Sesotho,which or variable,according to speakersex or languagespoken. might be characterizedas aspiratedand unaspiratedin accordancewith orthographicconventions. The unaspiratedclick has a meanVOT of ap- proximately40 mson theorder of voicelessaspirated plosives in English. The aspiratedclick hasa meanVOT of approximately120 ms, greater 1:32 than any plosivetype reportedby Liskerand Abramson[Language 47, 767-785 (1971)]. The Fo valueswere measuredpre- and post-clickin K3. Temporallocation of s,yllabic tonal target in standardChinese. both word•initial and word-medial clicks. Our data indicate a consistent Zongji Wu and Hongmo Ren (Phonetics Laboratory, Linguistics fall in Fo immediately prior to a click, regardlessof lexicaltone, presence Institute,Chinese Academy of SocialSciences, Beijing, People's Republic of word boundary, and the distinctive prosody of interrogatives.This of China ) perturbationof fundamentalfrequency may be indicative of an intrinsic The understandingof the dynamicnature of the syllablehas been propertyof click production. By contrast, measures of post-clickFo reveal deepenedby investigatingthe P center/stressbeat and segmentaltarget a lessregular pattern of microchangeboth across the silentinterval and concentration(Ren, 1986). This studyis concernedwith anotherfeature, duringthe initial glottal cycles at post-clickvowel onset; these phenomena the temporallocation of syllabictonal taxget in a tonelanguage, Chinese. areapparently influenced by lexicaltone. Chinesehas four syllabictones with the phonologicalforms H, LH, L, HL, respectively,realized by changes in F0. Ouracoustic data from three speakersshows that ( 1) in contrastto theanticipatory cross-syllable coar- 1:17 ticulationat the segmentallevel, the cross-syllabletonal transition is a carryover,(2) with a conflictingpreceeding tonal context, the temporal ![2. Interaction effectsof nasality and speakingfundamental frequency. locationof the only tonal targetin H and L and the first targetin LH and CarolineG. Henton (LinguisticsProgram, University of California, HL dependson the rateofF0 transitionstarting from the syllableonset, Davis, Davis, CA 95616) (3) with a harmonicprcceding tonal context,the first targetin LH and
S26 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America $26 HL is locatedvery closeto the consonantoffset or the vowelonset. The sian) on perceptualcategorization of ejectlyeuvular and velarstops in relationsbetween P center/stressbeat and tonal target in ease(3) callfor Thompson(a North AmericanIndian language). While theejectire man- furtherinvestigation. ner classis not representedin the phoneticinventory of eitherEnglish or Farsi,the uvular-velar distinction has a differentphonological status in the two languages.In English,the uvular-velarplace contrast is not a phoneroledistinction. In Farsi, this placedistinction is phonemiefor 1:47 voiced(nonejective) stops. To evaluatethe effectof this phonological statusdifference,, perceptual categorization ofmultiple natural exemplars K4. An acousticstudy of tonesandhi in Taiwanese.Hwei-Bing Lin of Thompson/k/and/•l/ (producedin the samevowel context) was ( Haskins Laboratories,270 Crown Street, New Haven, CT 06511-6695, assessedin both languagegroups. Subjects completed both an identifica- and Departmentof Linguistics,University of Connecticut,Storrs, CT tiontest and a name-identity(categorical) same-different discrimination 06268 ) test.Results are discussedas they addressthe role of phonetic(articula- toryfeature similarities between native and normatire phonologies in per- In Taiwanese,there are fivelong tones: high level,high falling,mid ceptionof nonnativespeech contrasts. [ Research supported by NINCDS falling, low rising,and mid level.These tones undergo sandhi changes and NICHD. ] when they do not precedea major syntatieboundary: mid levei•mid falling• highfalling • highlevel • mid level• low rising.The aim of this studywas to validatethese phonological observations by investigatingthe acousticproperties of tonesin sequence.Another issueof interestwas 2:32 whether tonal coarticulation occurs in Taiwanese. Six native speakers producedseveral repetitions of a sequenceof two syllables/si/and/do/ K7. Perceptualswitching in Spanish/Englishbilinguals. O. S. Bohn and in sentence-finalposition with all combinationsof tones.These utterances I.E. Flcge(Biocommunication, University of Alabamain Birmingham, wereanalyzed to determinethe average fundamental frequency (F o) con- Birmingham,AL 35294) toursof the fivetones on eachsyllable. The tonal contourson/do/sylla- bleswere highly similar to thoseobserved previously in isolatedsyllables. Elman etal. [J. Acoust. SOc.Am. 62, 971-974 (1971)] found that, Someslight perserveratory effects of the preceding/si/toneson the Fo whereasEnglish monolinguals heard short-lag stops as/b/, Spanishmon- onsetof/do/tones werefound. The toneson thepenultimate syllable/si/ olingualsheard them as/p/. Spanish/Englishbilinguals who were profi- underwentsandhi changes, as expected.No eoarticulatoryeffects were cientin Englishgave more "English" (i.e.,/b/) responsesin an English foundon/st/tones. However,the sandhitones differed in their Fo con- than Spanishperceptual "set." This studyfurther tested the hypothesis tours from thoseon the final /do/ syllable.[Work supportedby thatbilinguals may use two criteria in identifyingshort-lag stops. In ex- NICHD. l perimentI,/to/tokens spokenby Spanishmonolinguals with short-lag VOT werepresented along with Englishlong-lag/to/or Englishshort- lag/do/tokens. Englishmonolinguals and Spanishsubjects who learned Englishas adultslabeled the Spanish/to/stimuli as/to/(rather than /do/) moreoften in thecontext of English/do/( 73% ) than/to/(50% ). 2:02 Sinceboth groups showed the samepattern, this likely reflected a context KS. Prosodiecharacteristics and their controlin Japanesespeech under effectrather than a criterionshift. In experiment2, the Englishmonolin- varying speakingstyles. Masanori Miyatake and Yoshinori Sagisaka gualsdid notlabel the Spanish/to/stimuli as/to/more oftenin a Spanish (ATR InterpretingTelephony Research Laboratories, Twin 21 Building, than English"set." Whereasthe bilingualswho participatedin the Span- MID Tower,2-1-61 Shiromi Higashi-ku, Osaka 540, Japan) ish setfirst showeda 1256shift, thosewho participatedin the Englishset firstshowed no evidenceof a switchin criteria.It hasbeen hypothesized Two setsof Japanesespeech samples were analyzed to clarifythe ef- [ F!egeand Eelting, J. Acoust.Soe. Am. 83, 729-740(1988) ] thatSpan- fectsofspcaking styles on prosodic parameters. The firstset of 4068 isolat- ishspeakers who learn English in earlychildhood establish separate cate- ed word utterancesconsists of 308 differentJapanese words uttered in goriesfor Spanishand English/t/. Work isunderway to determineif such sevendifferent ways, i.e., normal,slow, fast, strong, weak, high, and low. subjectswill showa largerlanguage set effect than the adultL2 learners The second set of 1 l0 conversational utterances consists of 11 different alreadyexamined. [Work supportedby NIH.] Japanesesentences uttered in fourdifferent conversational styles, i.e., nor- mal,tender, restless, and irritated, and in onenormal reading style. These sampleswere uttered by two professionalnarrators and their prosodic parameters[fundamental frequency (Fo) , powerand segmentaldura- 2:47 tion] werecompared. The followingtendencies were found from the anal- ysis:( I ) For mostordinary speaking styles (all but"high") F o and power Kg. Nativeand normatire perception of voicingin English.Joann Fokes are varied together.Correlations of 0.81 --0.99 were measuredin these {Schoolof Hearingand SpeechSciences Ohio University,Athens, OH cases.(2) Fo can, however,be controlledindependent of power,when 45701) and Z. S. Bond(Department of Linguistics,Ohio University, subjectswere instructed to do so (e.g.,speaking style: high). (3) The Athens, OH 45701 } averagevalues and the dynamicranges of Fo patternsdiffer between speakingstyles. To realizethese tendencies for speechsynthesis by rule,a Naturallyproduced stop-vowel syllables (pea/bee, pie/bye, tea/dee, Fo controlmodel is proposed.Applying analysis by synthesisto all sam- tie/dye) wereselected to representvarious voice timing values from the plesin thefirst set at a time,the system parameters of thismodel could be prevoiced,unaspirated, and aspirated ranges. The setof tokensincluded derived.It is confirmedthat theF o patterncan be systematically genera- someintended voiced and intendedvoiceless stops produced with the tedby the proposedmodel for variousspeaking styles. samevoice onset timing (VOT) valueswithin the range + / -- 15 ms. Nativeand normatire English speakers identified three repetitions of each syllablein randomorder for a totalof 96 testitems. As expected,Ameri- can Englishlisteners judged all valuesin excessof 40 ms. VOT as voice- less,and all other valuesas voiced.In almost all cases,normatire listener responseswere predictableon the basisof their nativelanguage. Since 2:17 Chineseand Japanese contrast stops on thebasis of aspiration,listener K6. Cross-languageperception of a place-of-articulationcontrast: The judgmentswere almost identical to the American. On the other hand, infiueneeofnonspecifielinguisticexperience. Linda Polka(Department MalaysJanlisteners identified all voicelessand voiced stops as intended by of Psychology,University of SouthFlorida, Tampa, FL 33620) thespeakers; that is, listeners based voicing judgments on other properties ofthe syllables than voice timing values. The Arab judgments were almost Thereis somedebate whether the effectsof experienceon perception identicalto theAmerican for theaiveolar stops/t,d/but showed consid- of nonnativespeech contrasts reflect psychoacoustic, phonetic, or phone- erableuncertainty for the labialstops/p,b/. Bengalispeakers also fol- mic factors.The presentstudy addresses this issueby comparingthe per- lowed the American voicingpattern but showedsome confusion for formanceof nativeEnglish speakers and nativespeakers of Farsi (Per- wordscontaining the diphthong/ai/.
S27 J. Acoust.Soc. Am. Suppl. 1, VoL83, Spring1988 115thMeeting: Acoustical Society of America S27 3.•2-3:12
Break
3:12 3:42
K9. Voicing in French and English labial stops: A cross-language Kll. Formantonset duration: A eecondarycue for identifyingthe placeof perceptual study. Murray J. Munro and Terranee M. Nearey articulation in stop consonants. H. Garudadri, M.P. Beddoes (Departmentof Linguistics,University of Alberta, Edmonton,Alberta (ElectricalEngineering, University of British Columbia,Vancouver, T6G 2E7, Canada) British Cohimbia V6T 1W5, Canada), J. H. V. Gilbert, and A.- P. Benguerel(Audiology and SpeechSciences, University of British Measurementsof EuropeanFrench and Canadian English labial stops Columbia, Vancouver,British Columbia V6T 1WS, Canada) indicatethat the French/p/and English/b/categoriesare very similar, oftenexhibiting only small differences in voiceonset time. Two groupsof Formantonset duration (FOD) is suggestedas a possiblealternative English-speakinglisteners (one with someknowledge of French,the oth- to voiceonset time (VOT), to characterizethe placeof articulationin stop er without) wereasked to categorizea setof modifiednatural tokens from consonants. The FOD is defined as the duration from the start of the burst thesecategories as either/p/or/b/. The tokenswere taken from word- (denotingconsonantal release ) to the startof the formantstructure (de- initial stopsproduced in a sentencecontext. Components preceding the noting vowel onset). In many languages,VOT is the most salientcue releasebursts were removed, and the signalswere truncated at 68 ms. distinguishingbetween voiced and voicelessstop phonemes. Although Despitethe small difference in VeT betweenthe categories, most listeners VOT has been shown to increase in duration from bilabials to alveolars to wereable to reliablyseparate them at levelsabove chance. Analysis re- velars[Lisker and Abramson,Word 20; Zue, Tech. Rep. $23, MIT], yealedthat the listenersmay have relied primarily on VeT in making attemptsto useVOT as a cue to placeof articulation[ Kewley-Port,J. theirjudgments and that overallamplitude [C. J. Darwin and M. Pear- Acoust. Soc. Am. 73, 322-335 (1983)] were not successful.In a recent son,Speech Commun. 1, 29-44 (1982) ] may haveplayed a secondary study of stop consonantsusing the partially smoothedWignet distribu- role.These findings indicate that listenersmay be sensitive to smalldiffer- tion [Garudadriet al., J. Acoust.Soe. Am. Suppl. 1 82, S55 (1987)], encesbetween categories in their nativelanguage and analogous categor- FOD and F2 of the followingvowel were usedalong with the spectral iesin a foreignlanguage. shapeduring the burst,to resolvecontext dependencies due to eoarticula- tion effects. The FOD increases from bilabials to alveolars to velars and appearsto be relatedto the inertia of the articulators.Unlike VOT, FOD isalways positive and correspondsto the vocaltract ratherthan the source of excitation.Evidence from English,Telugu, and Frenchstop conson- ants supportingthese ideas is presented.[Work supportedin part by 3:27 NSERC, Canada.] KI0. Acoustic characteristics of tongue root position and vowel assimilationin Alcan.Susan Hess (Departmentof Linguistics,UCLA, 405 Hilgard Avenue,Los Angeles,CA 90024) 3:57
The purposeof this researchwas twofold:( 1) To find an acoustic K12. Prenasalizedstops and speechtiming. Ian Maddieson(Phonetics measure(or diagnosis)of tongueroot positionin a languagewhere a Laboratory,Department of Linguistics,UCLA, LosAngeles, CA 90024) differenceof tongueroot positionis systematicand significant,and (2) applythis measure to casesof vowelassimilation between vowels of con- Prenasalizedstops raise a numberof interestingquestions concerning fiictingtongue root positionsin orderto determinewhether tongue root therelationship between phonological units and timing in speech.Fijian is positionswere affected. Akan, a Kwa languagespoken in Ghana,exhibits a languagein which prenasalizedstops are incontrovertiblysingle seg- a form of vowelharmony controlled by the tongueroot in which the mentsfrom the phonologicalpoint of view.It hastwo stopseries, voiced vowelsin a particularword must be spoken with eitheran advancedor a prenasalizedand voicelessoral. All syllablesare openand no consonant retractedtongue root. The positionof the tongueroot [i.e., eitherad- clustersoc.cur. Hence, this is a particularlyappropriate language in which vancedor retracted;hereafter referred to as either + or - ATR (ad- to examinethese questions. Utterance-initial and word-media]prenasa- vancedtongue root) ] isdetermined by theword stem, and all affixesmust lizedstops at bilabial,a]veolar, post-alveolar, and velar places of articula- agreewith that tongueroot position. In thispaper, we will lookat data tion wereexamined in a setof wordsspoken by 11speakers representing from Kwawu, an Akan dialect,which has the followingvowels: [i,e,u,o] threemain varieties of Fijian.Acoustic duration was measured from onset are [+ ATR] vowels, while [l,e,o,•,a] are [--ATR] vowels. All of nasalmurmur to stoprelease (the stopcomponent of a prenasalized [ -- ATR] vowelshave variantswhen they occurin a syllablethat pre- stopis usuallyvery brief and was not measuredseparately). Initial prena- cedesanother word containingthe high [ + ATR] vowels[i,u]. Two saliz•l stopsare significantlyshorter than roedialones. Media] prenasa- claims have been made about these variants. One asserts that the variants lized stopsare comparablein durationto intervoca]iclaterals and voice- assimilatein tongueroot positionto the following[ + ATR ] voweland lessstops, but initial prenasalizedstops are significantlylonger than the become[ + ATR] vowelsthemselves, while the other arguesthat the initial nasalsin unstressedsyllables which occur in the data set.Vowel assimilationdoes not involvethe tongueroot but merelyraising of the lengthbefore prenasalized stops is comparableto that beforeother con- tongueblade. Resolving this dispute required that we firstdefine a mea- sonants.These results indicate that the roleof prenasalizedstops in the surethat wouldprovide a reliablediagnosis of tongueroot position. Only timing patternof Fijian is generallysimilar to that of other singleseg- acousticmeasures were consideredas other meanswere not practical. ments.In particular,they do not act like a consonantsequence. Five kindsof measureswere examined: formant frequency, formant band- width,duration, pitch, and amplitude of the firstthree harmonics. It was foundthat frequencybandwidth (more preciselythe bandwidthof the first formant) providedthe only reliablediagnosis of voweltype. The [ + ATR] vowelswere found to havea greatlynarrower first formant 4:12 whencompared with [ - ATR] vowels.This measurewas then usedto KI3. Differences in relative temporal positioning between initial examinethe behaviorof one of the [ -- ATR] vowels,the mid vowel [ el, consonantand vowel in syllables:Evidence from Chinese.Hongmo Ren in two contexts:before the high [ + ATR] vowels[i,u], and beforethe (PhoneticsLaboratory, Linguistics Institute, Chinese Academy of Social low [ -- ATR] vowel [a]. Bandwidth measurementsof e before the Sciences,Beijing, People's Republic of China) [ + ATR] vowels [i,u] do not show the narrower bandwidth found in [ + ATR] vowels,allowing us to preliminarilyconclude that tongueroot Juncturesand coarticulationsbetween syllable initial consonantand positionis not assimilated. vowelhave usually been described in termsof transitionsand target varia-
$28 J. Acoust.Sec. Am. Suppl. 1, VoL83, Spring1988 115thMeeting: Acoustical Society of America 828 tions.This study proposes that we also need to specifydifferences in rela- The presentexperiment was designed to discoverif this distinctionbe- tive temporalpositioning between these two majorsyllabic constituents. tweenthe homonymswas also retained in an isolatedenvironment; i.e., if The C +/iau/syllables in Chinesewere examined. The/iau/is a triphth- speakersdistinctively pronounced the homonymsin isolationand if lis- ongrhyme. Because rhyme internal formant pattern remains constant in tenerscould correctly identify them. FollowingFourakis and Iverson differentsyllables, the turningpoint in formanttrajectories can be used as (1984), the influenceof the speakers'attention to the purposeof the referencefor determiningthe temporalposition of the rhymeas a whole. experimentwas considered in theanalysis of theexperiment. Four pairsof In fact, we haveobserved the wholerhyme shifting along the time axis homonyms,of presumablyequal frequency in use,were recorded in two relativeto the initialconsonant boundary. Data fromsix speakers shows ways:first, subjects unaware of the homonymsorally responded to ques- thatsyllable initial consonant and rhyme exhibit different degrees of over- tions whoseanswers contained the homonymsas well as other decoy lappingand "repelling,"with a distanceranging from -- 15to 83 ms.The words;second, subjects orally read a list of the samehomonyms. The relationbetween these findings and the P centeror stressbeat measured in obtainedhomonyms were compared with respectto Fo value,duration, the sametokens is alsodiscussed. [Work supportedby NSF.] andpitch contour and then used for listeners'identification task. Both the acousticalanalysis and the resultsof the listeners'identification task provedthat onemorn homonyms were completely neutralized in an iso- latedenvironment regardless of the speakers'attention to the homonyms. [Specialthanks to JohnOhala, MasayoshiShibatani, Yoko Hasegawa, and MariscelaAmador. ] KI4. Analysisand prediction of voweldevoealization in isolatedJapanese words. Hisao Kuwabara and Kazuya Takeda (ATR Interpreting TelephonyResearch Laboratories, Twin 21 BuildingMID Tower,2-i-61 Shiromi,Higashi-ku, Osaka 540, Japan)
An analysisof voweldevocalization in Japaneseis presentedas an 4:$7 exampleto find generalrules for allophonicvariations of pbonemesin continuousspeech, i.e., to find quantitativerelations between phoneme K16. Delayedpitch fall in Japanese.Yoko Hasegawa(Department of environmentsand allophonic variations. A speechdatabase consisting of Linguistics,University of California, Berkeley, CA 94720} and about5000 common Japanese words spoken in isolationby a professional Kazue Hata (Speech TechnologyLaboratory, Santa Barbara, CA maleannouncer was used. The databaseused is partof a largescale speech 93105 ) databasethat is now being compiled and hand transcribed at severallevels of phoneticdetail. Based on the acoustic-phonetictranscriptions of the TheTokyo dialect of Japaneseis regardedas a prototypicalpitch ac- centlanguage where the location ofF0 fallis the only acoustic correlate of speechdata, factor analysis of the phoneticenvironment was performed for the devocalized vowels. The results reveal that the influence of the theaccent. Nenstupn 9 [Onsei-gakkaiKaihoo 121 (1966) ], however,re- portedthat F0 falldoes not always synchronize with an accented morn but immediatelyfollowing phoneme on devocalizationisthe greatest among maybe carried over to thefollowing mora. Sugito [ ShooinJoshi Daigaku the fourfactors examined. The secondmost important factor is the posi- Ronshuu10 ( 1972) ] foundthat nativespeakers perceived a pitchaccent tion of the accentnucleus. The automaticprediction of voweldevocaliza- on amora whenit wasfollowed by a fallingF0 contourand also that the tionis discussedusing the resultsof thisanalysis. F0 peakvalue in theaccented morn need not be higher than that in the followingmorn. Not wellinvestigated lhough is thecorrelation between thepeak location and the steepness of falling contour. The working hy- pothesisis that thelater the F0 fall occurs,the faster F0 drops.This delayedpitch fall is characterized interms of ( 1} theF0 peaklocation and K15. Pitch neutralizationin Japaneseone mora homonyms.Yuko (2 } thesteepness ofF0 contourcomputed in Hz/cs. The results from 560 tokensuttered by sevenspeakers of theTokyo dialect confirmed our hy- Mogami( 1431-1Toba Akashi, 673, Japan) pothesis.[Work partially supported by SIoan grant to BerkeleyCognitive Japaneseone morn homonyms in phrasesare distinguished by pitch. ScienceProgram. ]
TUESDAY AFTERNOON, 17 MAY 1988 ASPEN ROOM, 1:30 TO 4:50 P.M.
SessionL. Musical AcousticsI: Synthesisof the SingingVoice
Ingo R. Titze, Chairman Departmentof SpeechPathology and `4udiology,University of lowa, 127,4SHC, Iowa City,Iowa 52242
Chairman's Introduetio•--l:30
Invited Papers
LI. Computergeneration of singingby manipulationof speechparameters. Joseph P. Olive (Linguistics Department,AT&T BellLaboratories, Murray Hill, NS 07974) Thispaper describes a computer system for simulatinga singing voice by using parameters derived from naturalspeech. The system was constructed to manipulate parametrized speech to study prosodic and segmen- tal propertiesof speech.The system enabled a userto manipulatefundamental frequency, amplitude, timing,
S29 J. Acoust.Soc. Am. Suppl. 1, VoL83, Spring1988 115thMeeting: Acoustical Society of America S29 andspectral characteristics ofspeech sounds. While manipulations ofthe parameters ofnatural speech could easilyproduce "unreal" or "weird"sounds, simulating a singing voice with qualities approaching those of a naturalsinging voice was more difficult. The latter required careful manipulations ofthe fundamental frequen- cyand timing to imitate human singers. Examples ofboth types of sounds will be demonstrated. In an attempt to imitatenatural singing, the simulation of soloand choral singing, as well as contrasting singing styles of "classical"and "pop" singers is described.
2:05
L2. Singingsynthesis with modelsof phonatoryand articulatory kinematics. lngo R. Titze (VoiceAcoustics andBiomechanics Laboratory, Department of SpeechPathology and Audiology, University of Iowa, Iowa City, 1.4.52242 and Recording and Research Center, the Denver Center for thePerforming Arts, Denver, CO 80202 ) Singingsynthesis todate has been attempted primarily in theacoustic domain. In aneffort to learnmore aboutthe interactions between the source and subglottal and supragiottal resonators, a simulation strategy is employedthat computes pressures and flows on the basis of vocalfold and articulatory movement. Some aspectsofthe perception ofvocal registers are shown to be related to subgiottal resonances. In particular, the firstsubglottal formant (ca. 500 Hz) helpsto triggerthe involuntary chest-falsetto transition in untrained voices.Roundness in the medialsurface contour of the vocalfolds also helps to distinguishchest-falsetto and male-femaledifferences. Vocal tract constrictions in the vicinity of the glottis are shown to affectvocal quality, inparticular, the perception ofvocal ring. Frication inthe glottis and the vocal tract is computed onthe basis of criticalReynolds numbers and is an automaticconsequence of changes in subglottalpressure and airway constriction.Some examples ofsimulated sound will be given. [ Worksupported by NIH, GrantNo. NS 16320- o•.l
2:35
L3. Analysisand synthesisof the singingvoice and mnsicalapplications. Xavier Rodet (IRCAM 31, rue Saint-Merri, 75004 Paris, France) IRCAM (Iustitutde Rechercheet de CoordinationAcoustique/Musique) has been involved in sound analysisand synthesis for contemporary music creation for manyyears. Accurate analysis and high-quality soundsynthesis are thus extensively studied. Some of themethods that have been developed are described with particularattention to thesinging voice and speech. A powerspectral density estimate has been developed of thespeech signal, which is simultaneously very precise and computed on very short segments of thesignal in orderto follow rapid transitions such as those found in plosives.Spectral envelopes are represented in a flexible way:They are coded in termsof frequency, amplitude, and bandwidth of pairsof conjugate poles (formants). An algorithmhas been developed giving an harmonic/noisy decision in narrowfrequency bands covering the entirespectrum. This technique has greatly improved the sound quality. Several synthesis methods have been developedand compared, including formant waveform synthesis (analogous to parallelformant synthesis), additivesynthesis, and serial filter synthesis. Two types of synthesis rules have been studied. On the one hand, rulesconcerning phoneme coarticulation usingformant coded diphones, sophisticated algorithm smoothing formanttrajectories by taking into account three successive phoneroes, durations of phoneroes,speed of sing- ing,speed of transitions, and relative timing of formanttrajectories. On the other hand, a setof ruleshas been builtwith musical performance. Those rules calculate such features as vibrato excursion and rate, vocal effort, andformant locations according to fundamentalfrequency. Examples of synthesisby ruleare given showing differenttypes of voicesand styles plus excerpts from concert pieces.
3:05
IA. Singingsynthesis and composition. Gerald Bennett(Swiss Center for ComputerMusic, Oetwil am See, Switzerland ) Apartfrom its interestas a confirmationof experimentalhypotheses, singing synthesis can be of great importancein contemporarycomposition. The experienceof manycomposers with the programCHANP, developedat IRCAM, showsthat a soundsynthesis program based on a specificphysical model can serve as a generalpurpose synthesis tool, while remaining intuitively manageable. The particularstructure of the pro- gramCHANP offers the possibility of definingboth the local structure of thesound itself and the larger scale structureof the composition.Musical examples will illustrateboth vocal and nonvocal sounds taken from electroacousticcompositions.
3:35
[3. Synthesizingsinging. Johan Sundberg (Department of SpeechCommunication and MusicAcoustics, RoyalInstitute of Technology[KTH], S-10044, Stockholm,Sweden} An attemptto synthesizesinging is presented in whicha synthesizercontaining five cascaded formants is complementedby twonoise generators and fixed filters. The performanceis controlledby computersreading
S30 J. Acoust.Soc. Am. Suppl.1, VoL 83, Spring1988 115th Meeting:Acoustical Society of America S30 theinput notation and processing it with pronunciationand musical performance rules. Both solo and choral singingare synthesized. Various effects will bedemonstrated, such as timing of pitchchanges and consonants, coloratura,markato, timbre, legato, and multiple pitch singing. Also, some effects concerning musical expres- sionwill bepresented.
ContributedPapers
4:05 stimuliwere sinusoidally modulated at 5 Hz with frequencyextent of 0.25, 0.5, and0.75 semitones.Subjects were instructed to adjustrate/extent of L6. Is there a single vibrato waveform? Robert Maher and frequencymodulations until the differencebetween the standardand ad- JamesBeauchamp (Computer Music Project, School of Music, justablesignals was just noticeable.The ratejnd wasless than 5%, while Universityof Illinoisat Urbana-Champaign,2136 Music Building, 1114 frequencyextent jnd wasmuch smaller especially for thepulse stimuli. A W. Nevada, Urbana, IL 61801 } perceptualcue for the frequencyextent appeared to be rateof frequency High-fidelitysinging synthesis requires careful consideration of the changesrather than the extremefrequency values. A follow-upexperi- propertiesand characterof naturalvibrato. In manysynthesis methods, a ment indicatedlisteners' inability to matchaccurately nonmodulal•l sig- singlevibrato waveform is derived and applied equally to all of thepartials nal frequencyto the extremefrequency values of the modulatedsignals. of thesynthesized tone. The researchreported here tests the validity of Implicationsof thesefindings to vibratosinging will bediscussed. [Work this linear treatmentof vibratousing an analysis/synthesisprocedure. supportedby NIH. ] Thesinging tone/ah/is analyzedfor bass, tenor, alto, and soprano voices, providingtime-variant measurements of amplitude, frequency, and phase 4'•$ for eachof the partials.The tonesare reconstructedusing an additive synthesisprocedure. The importance ofvibrato waveform similarities and LB.Analysis, resynthesis, and modification of soundsignals with the help differencesbetween partials is evaluatedby resynthesisof the singing of wavelet transforms (i.e., time-and-scale representations). tonesfrom modifiedanalysis data. The modificationsinclude the useof R. Kronland-Martinet (Laboratoire de Mrcanique et d'Acoustique, the fundamentalpartial's vibrato waveform for all of the partials,the use C.N.R.S., 31, cheminJ. AIguier, 13402Marseiile Cedex 9, France) and of sinusoidalvibrato for all the partials,the elimination of amplitudevar- A. Grossmann(Centre de PhysiqueTh6orique, C.N.R.S.-Luminy Case iationsfor eachpartial, and variouscombinations. The analysisdata and 907, 13288 Marseille Cedex 9, France) examplesof theresynthesis will be presented.[ Work supported,in part, by an NSF Graduate Fellowship.] An attemptwill bemade to presentan up-to-datesurvey of resultson wavelettransforms of signals,with particularemphasis on examplesin- volvingsound. The main feature of a wavelettransform is that it associates to a signals(t) an appropriatefunction S(b,a) of twoparameters: a time b anda scaleparameter a. If it is requiredthat thecorrespondence s• S be 4'.2O linear,the definitionof a transforminvolves an "analyzingwavelet" g, whichhas to satisfysome conditions, and, given by S(b,a)----Ka -112 L7. The jnd's for rate and extent of frequency modulations. Xœs( t)g- [ (t -- b}/a]dt, g- is the complexconjugate ofg. This corre- Yoshiyuki Horii and gon Scherer (Campus Box 409, CDSS, The Universityof Colorado,Boulder, CO 80309and The DenverCenter for spondencehas a stableinverse. A transform$(b,a } isfully determined by its valueson a suitablediscrete grid of the (b,a) openhalf-plane. Wavelet the PerformingArts, Denver,CO 80202) transformsare close relatives of time-frequencyrepresentations, the fre- Vocalvibrato samples were spectrographically analyzed to determine quencybeing replaced by a scaleparameter. The aim will beto describe rangesof rateand extent of frequencymodulations. Subsequently, sinu- someof themathematical background, to givea descriptionof someexist- soidaland pulsesignals of 2-s durationat 261 or 522 Hz, modulated ingimplementations, and to displayan appropriateset of examples,in- within the measuredranges, were used as standardstimuli to deriverate volvingacoustical signatures in thefield of speechand music. There will andextent jnd. For the ratejnd, the standardstimuli were sinusoidally beauditive examples showing the possibilities of intimate modifications of modulatedat 4, 5, and 6 Hz. For the frequencyextent jnd, the standard soundby thesemethods.
TUESDAY AFTERNOON, 17 MAY 1988 GRAND BALLROOM A, 1:30 TO 4:20 P.M.
SessionM. Noise III: Airport Noise Compatibility Planning
Dennis G. Ossenkop,Chairman Federal•4uiation Administration, 1 7900Pacific Highmay South, Seattle, Washington 98056
Chairman's Introduction---l:30
Inuited Papers
1:35
M1. Airportnoise compatibility planning perspectives. Steven Starley (Noise Abatement Division, Federal AviationAdministration, 800 IndependenceAvenue, S. W., Washington,DC 20591) Air transportationismass transportation and its continuing growth has spawned a complex set of interre- latednoise and capacity problems. Airport proprietors, who are responsible for noisedamages, are turning to
S31 J. Acoust.Soc. Am. Suppl. 1. VoL83. Spring1988 115thMeeting: Acoustical Society of America the useof restfictionsto limit noisein adjacentcommunities. Use restrictions at oneairport affect system capacityand operations at other airports. When capacity limitations affect interstate cornmcrae, airports are caughtbetween their adjacent communities and the Federal government that cannot allow an undue burden on interstatecommerce. Airport noise abatement planning, which is effective for airport specific problems, may be a factorin thegrowth of national problems. Thus a nationalperspective must be applied to the solution of local problems.This papertraces the developmentof airportnoise abatement planning, discusses some newly emergingareas of interestat boththe local and national level, and presents a glimpse of somealternative futures.
2:05
M2. Attenuatingjet aircraft noisein single-familyresidences near major U.S. airports.Barney Myer (Facilitiesand Real Estate Department, Port of Seattle, SeaTac International Airport, P.O. Box 68727, Scatfie, WA 98168) Mostnoise attenuation of single-familyresidences near major U.S. airportshas been achieved by acquisi- tionand removal of homes.This noise remedy is not only extremely costly but it alsodisrupts social networks, undermineseducation and social institutions, lowers tax bases, and can generate subsequent land use problems. Acquisitionas a noiseremedy is limited to thoseareas where jet noiseis the worst. The thousands ofhomes that cannotbe purchasedbut endurenoise peaks of near 100dB mustbe addressedby otherremedies. SeaTac InternationalAirport and other airports are initiating acoustical insulation programs to lowerinterior noise levelsto acceptablelevels. An issueinvolved in thesedeveloping noise mitigation programs is how to setand achievenoise level goals that realistically make interior noise levels livable and are acceptable to the Federal AviationAdministration that funds up to 80% of programcosts. The issue involves answering questions such as:Are interior peak noise levels (SEL) tobe reduced low enough to allowuninterrupted communication (i.e., TV, telephone,conversation)? Is an averagenoise criterion (La,) sufficient?Is it possibleto standardize interiornoise goals when airports have widely divergent levels of activity?Noise remedy programs at SeaTat andother airports must address this basic issue to assure continued funding of their multimillion dollar efforts.
2:35
M3. The cooperativeapproach to noisecompatibility planning at Portland InternationalAirport. John P. Newell (Port of Portland, P.O. Box 3529, Portland, OR 97208)
The Port of Portland'scooperative approach to airportnoise compatibility planning has produced a plan that seeksto balancethe concernsof thecommunity with the needsof the airportand its users.This planhas reducedthe numberof residentswithin the Ld, 55 contourby 85%, withoutemploying costly operating restrictions.The success achieved is directly attributable to thecooperative process that involves the utilization of an advisorycommittee to helpidentify and resolveairport noise issues. This committee,consisting of all partieshaving an interestin theairport noise issue, has worked together since 1982 in structuringand imple- mentingthe noise abatement plan. This planis madeup of threeinterrelated programs: an operationalpro- gram,a landuse program, and a reviewand monitoring program. The operationalprogram includes specific flighttrack assignments, while the land use program requires sound insulation of someresidential dwellings, andadoption of city and county noise overlay zones. The review and monitoring program is the part of theplan that insuresthe ongoingcooperative process, and maintains compatibility with the community.
3:05
M4. The use of local opportunitiesin aircraft noiseabatement planning. Jon M. Woodward (Coffman Associate•, Inc., 1200 E. 104th Street, Suite 350, Kansas City, MO 64131 ) Aircraftnoise abatement planning often overlooks low-cost, easy to implementsolutions that take advan- tageof unique local conditions. The focus is too often on generic "cookbook" solutions that are applied to every airportin thesame way, e.g., curfews, aircraft type restrictions, or noisebudgets. For everyairport, careful analysiswill reveal unique conditions that provide opportunities toabate noise. These include airport operating schemessuch as preferential runway use programs, rerouting of trafficalong corridors of compatible land uses, andthe useof comparativeaircraft versus background noise levels to directtraffic flows. Preferred aircraft operatingprocedures are selected based upon local underlying surface conditions of land use or terrain.Limita- tionson aircraftoperation through imposition of locallegislation can result in noisereduction in specialized cases.Construction of newfacilities most often is drivenby capacityneeds as well asnoise abatement. Exam- plesof theeffects of usinglocal opportunities for noiseabatement are presented for eachtype of measurefrom airportnoise compatibility planning projects at Portland,Seattle, Phoenix, Anchorage, Indianapolis, Baton Rouge,New Orleans,Chicago, Omaha, Little Rock, and Reno.
$32 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115th Meeting:Acoustical Society of America 832 3:35 MS.Use of seasonal Ld.for airport noise compatibility planning. Andrew S. Harris(Harris Miller Mille• & HansonInc., 429 Marrctt Road,Lexington, MA 02173) Thedifferences between Ld, fromyearly average operations and Ld, frombusy day or seasonal operations areexamined. Civil and military airports are considered. For civil airports, the issue of seasonalityisaddressed asit relatesto theFederal Aviation Administration's requirements under FAR Part 150.For bothcivil and militaryairports, current examples are discussed. The environments of civil and military airports are also examinedin thecontext of EPATask Group 3's recommendations about seasonality ["Impact Characteriza- tionof NoiseIncluding Implications of Identifyingand Achieving Levels of Cumulative Noise Exposure," for EnvironmentalProtection Agency Aircraft/Airport Noise Report Study, Task Group 3, Henningyon Gierkc, Chairman,Washington, I June 1973. ] Methodsfor developingseasonal noise exposure arc examined.
ContributedPaper
4:O5 sourceof disturbanceto parkvisitors. Under such quiet background con- M6. Measurementand analysistechniques for determiningacoustical ditions, enroute aircraft are often audible for extended durations. In the impactsfrom enrouteaircraft in very quietsettings. Robin T. Harrison NationalPark setting,it hasbeen found that eventhe detectionof intru- (1275 North Indian Hill Boulevard,Claremont, CA 91711) and Paul sivesound can result in annoyance.Numerous and continuing complaints H. Dunholter(Mestre Greve Associates,280 NewportCenter Drive, havebeen generated by overflightsthat wouldnot be expectedto havea Suite230, NewportBeach, CA 92660) significantimpact based upon traditional aircraft noise assessment crite- ria alone.This papersummarizes a NationalPark Servicefunded study The National Parks of the United States have been set aside for the that addressesvarious techniques for the measurementofenroute aircraft publicenjoyment of pristinesettings with minimalimpacts from common noisein thesequiet settings.These techniques include A-weighted and urbandisturbances. Ambient sound levels in thesesettings are generally detectability metrics for both the ambient and aircraft noise. Methods very low, and have been measuredbelow 20 dBA. Aircraft noisefrom includingthe effectsof the durationof the overflightand the numberof enroutehigh altitude aircraft and sight-seeingaircraft have becomea events are also reviewed.
TUESDAY AFTERNOON, 17 MAY 1988 GRAND BALLROOM B, 1:30 TO 4:30 P.M.
SessionN. PsychologicalAcoustics II: Masking, Modulation and Modulation Masking
ChristopherW. Turner, Chairman Departmentof CommunicativeDisorders, Syracuse University, 805 S. Crouse•tvenue, Syracuse, New York 13210
ContributedPapers
1:30 1:45
N1. Frequencyselectivity in focusedattention as demonstratedby a N2. Combined masking under conditionsof high uncertainty. Donna multiprobe method. Huanping Dai and Bertram Scharf (Auditory L. Neff, Walt Jesteadt,and Brian P. Callaghan (Boys Town National PerceptionLaboratory, 413 MU, NortheasternUniversity, Boston, MA Institute, 555 N. 30th Street, Omaha, NE 68131 ) 02115)
A multiprobeparadigm, patterned after the probe-signalmethod of Simultaneousmaskers composed of a few sinusoldsscattered over a Greenbergand Larkin [J. Acoust.Soc. Am. 44, 1513-1523(1968)], led widefrequency range produce large amounts of maskingif the frequency observersto focusattention on a primaryfrequency in a detectiontask compositionof the maskersis changedwith eachstimulus presentation. (2IFC). In eachblock of 64 trials,the signalwas presented 48 timesat the This maskingappears to be producedprimarily by morecentral processes primaryfrequency and twice at eachof eightprobe frequencies, four high- drivenby stimulusuncertainty. In this study,the maskingproduced by er andfour lower than the primary. By presentingmany probes in a block combiningthese maskers with broadbandnoise is examined.Growth of insteadof onlyone as in mostprevious work, across-blockvariability was maskingfunctions for broadbandnoise and for multicomponentmaskers avoided.It alsomade it unlikelythat observersfocused on frequencies with 2, 6, 10,50, and 100components were used to selectlevels for individ- other than the primary.Results from four observerstested in separate ual maskersthat produced10, 20, 30, or 40 dB of masking.Multicompon- sessionson primariesat 250, 500, 1000,and 2000 Hz revealeda consistent ent and noisemaskers were then presentedin equatedand unequated dropfrom 90% detectionof the primaryto chancedetection of probes combinations.The averageamount of additionalmasking beyond a power more than half a critical band away from the primary. At a primary of summationmeasured for equatedconditions decreased from around12 to 4000 Hz, three observersshowed the samekind of selectivity,but two 2 dB as the numberof componentsin the multicomponentmasker in- othersshowed less. Generally, informed and naiveobservers performed creasedfrom 2 to 100.The data are well fittedby Lutfi'smodel for com- muchthe sameway. Since practice also mattered little, a sharpfilterlike bined simultaneousmaskers, with exponentsthat approach1.0 as the functioncould be measuredin under 1000 trials. [Work supportedby numberof componentsincreases to moreclosely approximate broadband NIH.] noise.[Work supportedby NIH and AFOSR.]
S33 J. Acoust.Soc. Am. Suppl. 1, Vol. 83, Spring 1988 115th Meeting:Acoustical Society of America S33 2:00 from I dB in the burstcondition to 7 dB for a 450-msfringe, due to a greaterimprovement in detectabilityin the correlatedcondition com- N3. Changes in the masked thresholdsof brief tones producedby paredto the uncorrelatedand all-uncorrelated conditions. When the cue precedingbursts of bandpassand notched noise. Robert P. Carlyon bandswere gated before the masker,the averageCMR waslarger (4-6 (MRC Instituteof HearingResearch, University Park, Nottingham NG7 dB) than when the maskerwas gatedbefore the cue bands(2-4 dB). 2RD, England) Thesedifferences in improvementwith increasingfringe duration may be Thresholdswere measuredfor 5-ms l-kHz tonesmasked by synchro- attributableto differencesin neuraladaptation in the correlatedand un- nousbursts of noisecontaining a spectralnotch centeredon the signal correlatedconditions. [Work supportedby NINCDS Grant NS15895.1 frequency.These thresholds were reduced by priorexposure to a 200-ms burstof a "primingstimulus," which had the samespectral shape as the masker.This maskingrelease was greatestfor notch widths extending between20%-30% on eitherside of the signalfrequency, and was absent when the maskerand primer containedno notch. A smallermasking 2:30 releasecould be obtained with primersconsisting of only the lower band of a notchednoise masker, and, to a lesserextent, of thehigher band alone. A NS. Comodulationmasking release (C•R) as n function of masker primer consistingof a narrowband of noisecentered on the signalfre- bandwidth, signal duration, nnd modulator b•ndwidth. Gregory quencyproduced an increasein masking,which could not be attributed to P. Schooneveldtand Brian C. J. Moore (Departmentof Experimental forward maskingof the tone. The effectsof all of theseprimers were Psychology,University of Cambridge,Downing Street, Cambridge CB2 independentof the 30-dB raffgeof primer levelsstudied, ruling out an 3EB, England) explanationin termsof peripheraladaptation. The resultsare consistent with the presenceof an activeneural mechanism that enhancesthe inter- Thresholdswere measuredfor a 2000-Hz signal maskedby contin- nal representationof newlyarriving energy in a previouslyunstimulated uousnoise varying in bandwidthfrom 50 to 3200 Hz in l-oct steps.For frequencyregion. Temporal parameters of the maskingrelease were also randomnoise maskers, thresholds increased with increasingbandwidth studied. up to 400 Hz and thenremained approximately constant. When the mask- er wasamplitude modulated by a low-passnoise, so as to producecoher- ent envelopefluctuations across frequency, thresholds decreased as the maskerbandwidth was increased beyond 200 Hz, givinga CMR. For a 400-mssignal duration, the CMR for maskerbandwidths greater than 2:15 400 Hz increased from 2.4 to 12.3 dB as the modulator bandwidth was N4. Comodulationmasking release with delayedsignal onsets. Beverly decreasedfrom 400 to 12.5Hz in l-oct steps.For modulatorbandwidths A. Wright and Dennis McFadden (Department of Psychology, of 50 Hz or less,a releasefrom maskingof 3.5 to 7.3 dB occurredeven for Universityof Texas,Austin, TX 78712) maskerswith bandwidthsof 50 and 100 Hz, lessthan the critical band- width at 2000 Hz. For a modulator bandwidth of 12.5 Hz, the CMR The detectabilityof a 1250-Hz, 50-mssignal was assessed in a como- decreasedfrom 12.3to 5.3 dB asthe signalduration was decreased from dulation maskingrelease (CMR) setting.One 50-Hz noiseband (the 400 to 25 msin 1-octsteps. When the signal duration was less than 100ms, "masker") was centeredat 1250 Hz; other noisebands were centeredat there wasno releasefrom maskingfor maskerbandwidths less than 400 850, 1050, 1450, and 1650 Hz (the "cue" bands). The maskerand/or cue Hz. The resultssuggest that, for maskerswith fluctuatingenvelopes, bandswere gated prior to theonset of thesignal by an amount("fringe") across-channelcomparisons can reduce signal thresholds even for short that wasvaried across blocks of trials.The noisebands and the signal signals,but an extrawithin-channel process can produce a releasefrom alwaysended together. The temporalenvelopes of thenoise bands were all maskingfor longsignals. This secondprocess may reflectthe abilityof correlated,all correlatedexcept for the maskerband, or all uncorrelated. subjectsto detecta changein the statisticalproperties of the envelopeof When all of the noisebands were gated synchronously, the CMR grew the stimuluswhen the signalis addedto the masker.
2:45-3:00
Break
3:00 nentshave the same depth of modulationas the on-frequency component. [Researchsupported by AFOSR.] N6. Signal threshold as a function of the relative modulation depth betweenon-frequency and flanking masker components.John H. Grose and JosephW. Hall, III (Division of OtolaryngologyHead and Neck Surgery,School of Medicine,University of North Carolina,Chapel Hill, 3:15 NC 27514) N7. Gap detectionin a narrow-bandnoise with either a comodulatedor a It is apparent that the mechanismunderlying comodulation masking noncomodulatedflanking band. John H. Grose and JosephW. Hall, III release(CMR) relies on the existenceof temporal fluctuationsin the (Division of OtolaryngologyHead and Neck Surgery, School of maskerenvelope. This raisesthe questionof how much fluctuationis Medicine,University of North Carolina,Chapel Hill, NC 27514) necessaryto facilitateCMR. The presentexperiment addresses this ques- tion by measuringpsychometric functions relating signal threshold to Comodulationmasking release (CMR) suggeststhat the auditory depthof flankerband modulation using sinusoidally amplitude-modula- systemis sensitiveto across-frequencydifferences in modulationpattern. tedpure tones as masker components. In thefirst set of conditions,the on- This raisesthe questionof whetherit is assensitive to modulationdiffer- frequencycomponent had a constantmodulation depth of 100%, while encesdue to the absenceof activity (a silentinterval ) asit is to the pres- the depthof the flankingcomponents was varied between 100% and0%. enceof additionalactivity (a signal).If so,gap detection in a narrow-band As expected,signal threshold improved monotonically as depth of ranker noisewould be expectedto be better in the presenceof a comodulating band modulation increased. In the secondset of conditions, the on-fre- flankingband than in the presenceofa noncomodulatingflanking band. quencycomponent had a constantmodulation depth of 63%, while the The presentstudy was designed to testthis hypothesis. Gap detectionwas depthof the flankingcomponents was either 100%, 63%, or 0%. Results measured in a 30-Hz-wide narrow-band noise centered at 1 kHz. A second to datesuggest that signalthreshold is lowestwhen the flankingcompo- 30-Hz band of noise,centered at either 500 Hz or 1.5 kHz, was then added
S34 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S34 that was either comodulated or noncomodulated with the l-kHz band. giventhe appropriate choice of temporalwindow parameters. [ Work sup- While gapdetection deteriorated with theaddition of a secondnoise band, portedby AFOSR and NIH. ] it appearedto do somore for the noncomodulatedcase than for the como- dulatedcase. [ Researchsupported by AFOSR.]
4:00
NI0. Detecting amplitude modulation of sinusoidal carders. William A. Yost and Stanley Sheft (Parrely Hearing Institute, Loyola 3:30 University,6525 N. SheridanRoad, Chicago,IL 60626)
N8. Cross-channel effects in amplitude-modulation detection. Listenerswere asked to detect sinusoidalamplitude modulation StanleySheft and William A. Yost (Parmly HearingInstitute, Loyola (SAM) of onecarrier tone (the targetcarrier) in thepresence of a second University,6525 N. SheridanRoad, Chicago, IL 60626) carriertone (the maskingcarrier) that alsohad SAM. The depthof mod- ulationrequired to detectthe presenceof SAM of the targetcarrier was Sensitivityto low-frequencysinusoidal amplituc'te modulation (AM) measuredas a functionof the differencein the frequencyof amplitude at a single-componentfrequency of an equal-amplitudetonal complex modulationfor thetwo carriers(modulator frequencies ranged from 4 to wasinvestigated. All maskercomponents of thecomplex were modulated 1{30Hz), thefrequency separation between the two carriers(200- to 3000- with a fixeddepth at the samefrequency as the probe.The phasebetween Hz separation),and the phaseof the sinusoidalmodulator of the masker probeand maskerenvelopes was varied acrossconditions. For a two- carrier relativeto the phaseof the sinusoidalmodulator of the target componentcomplex, the changein performanceas a functionof frequen- carrier.In half of the conditions,the frequencyof the targetcarrier was cy separation( A f ) dependedon theenvelope phase relationship. With a greaterthan the frequencyof the maskercarrier, while, for the other half 2-kHz probecarrier, the slopesof the AM thresholdversus A f functions of the conditions,the frequencyof the targetcarrier wasless than that of are more gradual than would be predictedby envelopeinteraction in a the maskercarrier. The data will be discussedin termsof the strategies single-frequencychannel. Consistent with AM detectionmodels incor- employedby the auditory systemin processingtemporal modulation of poratingbroad prodetectionfiltering, performance may be affectedby complexsounds. The data indicatethat the auditorysystem uses a wide- detectionof changesin the overallmodulation pattern. When the probe bandmode for processingtemporal modulation, such that thereis a great andmasker envelopes differ, adding masker components at thefringes of dealof interactionacross widely separate frequency channels. [ Work sup- the two-componentcomplex can lead to an improvementin probeAM portedby the NINCDS.] detection.This resultsuggests that the detectionof the modulationpat- tern of an individualcomponent in a tonal complexmay be aidedby enhancingcross-channel differences in modulation.
4:15
NIl. Modulation masking patterns. Sid P. Bacon, D. 3:45 WesleyGrantham, and Luann E. Van Campen(Division of Hearing and Speech Sciences,Vanderbilt University School of Medicine, N9. Complexsound discrimination: Predictions of the EWAIF model.L. Nashville, TN 37232) L. Feth, L. J. Stover,and R. A. Gcrrcn (Departmentof Speech- Modulationthresholds for sinusoidallyamplitude-modulated (SAM) Language-Hearing,University of Kansas,Lawrence, KS 66045) broadbandnoise were obtained for modulationfrequencies from 2 to 512 The envelope-weightedaverage of instantaneous frequency Hz usinga two-interval,forced-choice adaptive procedure. The noise car- (EWAIF) modelof auditoryperception has been successfully applied to der wason continuously throughout a block of trials,and was modulated simple,two-component signals, simultaneously amplitude- and frequen- for 500 ms in one of the two observation intervals. Thresholds were ob- cy-modulatedtones, and evento the complexsignals used in the early tainedin quietand in thepresence of a SAM broadbandnoise masker. In "profileanalysis" work. The EWAIF modelpredicts performance from a the maskingconditions, the samenoise carrier, presented at a spectrum calculationof the envelope-weightedaverage of the instantaneousfre- levelof 15 dB SPL, wasused for the signaland the masker.The masker quencyfluctuations inherent in almostevery complex periodic sound. was modulated in both of the 500.ms observation intervals. The modula- Whenthe modelis in error,it generallypredicts better performance than tionfrequency of themasker was 4, 16,or 64 Hz; itsmodulation depth our listenersachieve. A revisedversion of theEWAIF modelincorporates (m) was0 (no modulation),0.5, or 1.0.For a givenmasker modulation a temporalprocessing window. The revisedmodel was tested by compar- frequency,the modulationmasking patterns generally were bandpass, ing its predictionswith listeners'performance in a frequency-changedis- with thegreatest amount of maskingoccurring when the signal and mask- criminationtask. The experimentrequires the listenerto distinguisha er modulationfrequencies were the same.With a few consistentexcep- smoothfrequency glide from a discrete,multistep transition over the same tions,there was a monotonicrelation between masker modulation depth trajectory.The listeners' ability to distinguishthe glide from the multistep andamount of masking(the greater the modulation depth the greater the transition,in a 2IFC task, decreasedto chanceas the number of steps amountof masking).These data indicate a tuningof the auditorysystem increased.The EWAIF modelperformance follows that of the listeners, for thedetection of modulation.[Work supportedby NIH.]
S35 J. Acoust.Soc. Am. Suppl.1, VoL 83, Spring1988 115th Meeting:Acoustical Society of America S35 TUESDAY AFTERNOON, 17 MAY 1988 WEST BALLROOM B, 1:30TO 4:50 P.M.
SessionO. UnderwaterAcoustics II: StochasticVolume and BoundaryScattering II
John McCoy, Chairman CatholicUniversity, Washington, DC 20064
Chairman's lntroductian--l:30
Invited Papers
1:35
O1. Imagingthrough random med'a. Mark J. Beran (Schoolof Engineering,Tel Aviv University,Ramat Aviv, Israel) The distortingeffects of randommedia on imagingare discussed.Random phase and randomintensity fluctuationswill betreated and the importanceof isoplanicitywill be emphasized.Methods of processingto overcomethe distortion in conventionalimaging will beconsidered, and the advantages of speckleinterferome- try and intensityinterferometry will be mentioned.
2:00
02. Rough interface scattering,David H. Berman, Dalcio K. Daeol, and John S. Perkins (U.S. Naval ResearchLaboratory, Code 5160, Washington DC 20375) Recently,there has been a greatdeal of workon scatteringby roughsurfaces. Most of thiswork is aimed at describingscattering from the seasurface, and so Dirichlet boundary conditions are used. However, there are situationsin whichenergy transmission across a boundaryis important, such as in scatteringfrom theseafloor. Hence,scattering by roughacoustically penetrable surfaces is considered.Methods for calculatingsuch scat- teringwill be brieflyreviewed. Results obtained by the "point-matching"method will be presentedfor fluid- fluid and fluid-solidinterf•.es, both deterministic and random.Solutions of theseproblems yield forward and backscatteringstrengths of the seafloor,ice canopies,as well as the seasurface. In addition,it is possibleto determinethe probabilitydistribution of scatteredenergy. These exact results will alsobe usedto evaluate variousapproximation schemes.
2:25L 03. Chao• in underwater•coustics. F. D. Tappert, M. O. Brown (Applied Marine Physics,RSMAS, Universityof Miami, Miami, FL 33149), D. R. Palmer,and H. Bezdek(NOAA/Atlantic Oceanographic andMeteorological Laboratory, Miami, FL 33149) The problemof predictingsound propagation in range-dependentocean environments has been investigat- ed, in whichit is sopposedthat the environment(volume and/or boundary)varies smoothly in rangeand is exactlyknown with arbitraryprecision. Although this problem as stated is deterministicand not intrinsically stochastic,it hasbeen discovered from numericaland analytical studies of physicallyrealistic examples drawn from deepocean propagation, shallow-water propagation, and surfaceduct propagation,that ray path solu- tionsexhibit "classical chaos," namely, unpredictable and stochastic behavior. Ray pathsare found to havea continuousspectrum characteristic of noisystochastic processes, and ray pathsare found to havean exponen- tially sensitivedependence on initial conditionsand environmental parameters characteristic of chaoticpro- cesses.This phenomenonof chaosin underwateracoustics is causedby the exponentialproliferation of catas- trophes(caustics) due to theloss of controlimplied by the nonseparabilityof variablesin theeikonal equation. As a consequence,even when the oceanenvironment is knownexactly, there exists a "predictabilityhorizon" that limitsthe rangeto whichacoustic fields can be predicted.
ContributedPapers
2:50 limitsof geometricaltechniques. Pulses (2, 4, 8, and 16kHz) travelinga 6-km path from sourceto receiverin the BeaufortSea were receivedat a 04. The limits of sourcelocalization due to stochasticocean proce•. 150-m verticalarray during the AIWEX acoustictransmission experi- Terry E. Ewart, Robert P. Porter, Juraj Dworski, and Stephen ment, A.ATF_• Pulse travel timesestimated using the BeiI-Ewart multi- A. Reynolds(Applied PhysicsLaboratory, University of Washington, pathseparation algorithm provide singie-path travel time accuracies of a Seattle,WA 98105) few p seconds.Treating the sourcelocation as unknownallows for the The locationof a sourcefrom acousticmeasurements may be estimat- studyof theprecision of variousmethods. Ray trace methods were imple- ed in severalway• A knownsource/receiver geometry is usedto studythe mentedto backpropagatethe travel time estimatesto a sourcelocation
S36 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115th Me(fling:Acoustical Society of America S36 usingthe meansound-speed profile. A simplesource location estimator enceeffects. Numerical simulations will be presentedthat examinethe basedon least-squares fitting of the measured to thepredicted travel times placementof receiversfor optimum signal detection at theselow frequen- providesa classical example of a globalversus local optimization problem cies.[ Work supportedby ONRANORDA.] whenthe array angle is included in theminimization. Methods have been developedto circumventthe optimization problem and range errors of a few centimetersand deptherrors of a few metersresult. The methods, 3:50 resultsshowing the effectsof the medium,and a discussionof further 08. Numericalmodeling of the scatteredacoustic field from elasticice. refinementsare presented. [Work supportedby ONR.] J. Robert Fricke (MIT/WHOI JointProgram, MIT, Cambridge,MA 02139), Ralph A. Stephen(Woods Hole OceanographicInstitution, WoodsHole, MA 02543), and Arthur B. Baggeroer(Departments of 3:05 Oceanand ElectricalEngineering, MIT, Cambridge,MA 02139) 05. Chaotic behaviorof ray trajectoriesin a range-dependentocean A two-dimensionalelastic finite difference modeling program written environment. D. R. Palmer (NOAA/AOML, 4301 Rickenbacker formarine geophysical applications [ M. E. Doughertyand R. A. Stephen, Causeway,Miami, FL 33149),M. G. Brown,F. D. Tappert(RSMAS/ J. Acoust.Soc. Am. 82,238-256 (1987) ] hasbeen modified to studythe AMP, University of Miami, Miami, FL 33144), and H. F. Bezdek ice scatteringproblem. The modifiedcode was checked by comparing (NOAA/AOML, 4301Rickenbacker Causeway, Miami, FL 33149) modeledplane-wave reflection coefficients at variousangles of incidence with the analyticalsolutions provided by Zoeppritzequations. These tests It hasbeen demonstrated that ray trajectoriespropagating in a range- werecalibrated for a fiat interfacewith homogeneoushalf-spaces of water dependentocean environment can exhibit chaotic behavior. A particular- andice and resulted in favorablecomparisons. Subsequently, various ice ly simplesound-speed model was considered consisting ofthe Munk refer- models were run to observe the scattered acoustic field. These models eneesound-speed profile to whichis addeda smallrange-dependent, includeflat ice, flat ice with a singlekeel, and ice with randomlyrough deterministicperturbation having an harmonic dependence onrange and surfaces(top and bottom) derivedfrom differentprocess models (e.g., decreasingexponentially with depth. Chaotic ray trajectories were identi- Gaussianand Poisson). With thisprogram, controlled numerical experi- fiedfrom an examinationof Poincar6sections and power spectra. The mentscan be performedfor singlerealizations of iceroughness including sensitivityof chaotictrajectories to initialconditions and the consequent full elasticbehavior of the ice whichis usuallyneglected in scattering implicationsfor predictability were investigated by considering the evolu- theory.Such a capabilitycan be usedas an analysistool for evaluating tionwith range of a bundleof rays that initially occupy a verysmall region scatteringapproximations (say, the Kirchhoffapproximation). In addi- in phasespace. The largestLyapunov exponent was determined by con- tion,this modeling capability can be used as the synthesis step of a scatter- sideringthe spreadingof the bundle.Since the ray equationsdefine a inginversion program where ice roughness parameters are to beestimated nonautonomousHamiltonJan system with one degreeof freedom,our from a measurementof the scatteredfield. [Work supportedby ONR resultscan be understood in termsof recentcontributions to the studyof classical chaos. Arctic Acousticsprogram.]
4:05 3:20 09. Images of waves through random media by numerical simulation. 06. Chaos in shallow-waterpropagation. F. D. Tappert and M. StanleyM. Flatt• andJan Martin (UniversityofCalifornia,SantaCruz, G. Brown (Applied Marine Physics/RSMAS,University of Miami, CA 95064) Miami, FL 33149) Usingthe SDSC/CRAY, an extendedrandom medium has been mod- Soundpropagation in downwardrefracting shallow-water environ- eledby a setof two-dimensionalthin phase-changingscreens with phase mentsis considered,in whichit is supposedthat the waterdepth varies powerspectral densities having various power-law indices and innerand smoothlyin rangeand is exactly known with arbitraryprecision, and the outer scales.The strengthof the indexof refractionin the mediumhas bottomis perfectlyreflecting at shallowangles. After a few additional beenvaried. At smallstrength, the propagationis unsaturated, and the assumptionsare used to simplifythe physics,it wasfound that the equa- randomintensity patterns at the finalscreen are dominatedby the effects tionsfor raypaths (RBR) undergoingmultiple bottom reflections reduce of the small-scaleparts of the medium.As the strengthof the medium to the "standardmap" of chaostheory, where the "stochasticityparam- fluctuationsis raised,the imagesof intensityat the final screenshow eter" K is four timesthe ratio of boundarycurvature to ray curvature. strikingpatterns, with causticsforming and interferencefringes near the Typical valuesof K greatlyexceed the critical value,K c=0.97, above causticsbeing visible. For evenlarger medium strength, regions of high which"global chaos" is knownto prevail.Thus ray pathsin thesedeter- intensityassociated with many causticsclump into large-scalepatterns ministicenvironments typically exhibit chaoticbehavior, namely, ray that are relatedto the large-scalestructure of the medium.Intensity spa- pathshave an exponentiallysensitive dependence on initial conditions. tial spectraand variancewill be given,but it shouldbe emphasizedthat the observedpatterns are not well characterizedby a simple,random- phase.realization of a powerspectrum. Some comparisons with experi- 3:35 mentsin laser-beampropagation through the atmospherewill be given.
07. Applicationof the finite elementmethod to simulatingVLF/ULF under-iceacoustic propagation and scattering.Stanley A. Chin-Bing 4:20 ( NavalOcean Research and Development Activity, Numerical Modeling Division,NSTL, MS 39529-5004)and Joseph E. Murphy (Department O10. Field and numerical volume scattering experiments.Terry of Physics,University of New Orleans,New Orleans,LA 70148) E. Ewart and StephenA. Reynolds(Applied PhysicsLaboratory and Schoolof Oceanography,University of Washington,Seattle, WA 98105) A full-waverange-dependent ocean acoustic propagation and scatter- ing modelbased on the finiteelement method [J. E. Murphyand S. A. Until recently,the historyof comparingocean acoustic volume scat- Chin-Bing,J. Acoust. Soc. Am. Suppl.1 82, S61 (1987) ] hasbeen applied teringexperiments with theorybore scant resemblance to the traditional toa shallow-waterunder-ice scenario at verylow to ultra-lowfrequencies scientificmethod. However, vastly improved agreement between predict- (VLF/ULF) and at extremeclose range. The modelhas the advantage ed and observedintensity correlations is now seen.It is clear that two that it includesthe coupledeffects of the full-waverange-dependent for- elementsof the researchhave played vital rolesin reachingthe current ward propagatedand backscatteredfields. (However, the madeldoes not understanding.The first is the developmentof theoriesthat includethe yet includethe effectsof shearwaves.) The under-icesurface was simulat- parametrizationsand approximations appropriate to the oceancase. The ed by a fractallikeinterface superimposed on an ice keelrepresentation. secondis the characterizationof the deterministicand stochasticpro- At very low frequencies,the acousticwave can penetratethe ice-water cessesof the oceanthat giverise to the observedfluctuations. An overview interfaceand undergoreflection at the air-ice interface.The reflected, of thoseprocesses relevant to the study of oceanvolume scattering is refracted,and backscattered waves couple to produceinteresting interfer- presented,followed by a discussionof whereattention needs to be focused
S37 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115thMeeting: Acoustical Society of America S37 in orderto expandour understanding of volumescattering. Results from The coherentpressure field of an acousticharmonic point source is the 1971 experimentat Cobb Seamount,the 1977 Mid-Ocean Acoustic graduallytransformed into an incoherentpressure field as a functionof TransmissionExperiment, the 1985AIWEX AcousticTransmission Ex- propagationdistance through the oceanicinternal wave field. A receiver periment,and numericalexperiments provide the frameworkfor the dis- locatedin the weakscattering region Would measure a fluctuatingacous- cussions.[Work supportedby ONR.] tic pressurewith a Gaussianlydistributed logarithmic amplitude and phase.The consequenceis a biasedestimate for the conventionalmaxi- mum likelihoodestimate of pressureor intensityand subsequentlya bi- 4:35 aseddetection criteria. It is proposedthat detectionperformance may be improvedin this caseby logarithmicallytransforming the pressureprior O11. Acoustic detection in the weak scatteringregion of an oceanic to detectionprocessing. An analysisand discussionwill be presentedus- internalwave field. BruceJ. Bates(Naval UnderwaterSystems Center, ing the Garrettand Munk internalwave field model. [ Work supportedby NewportLaboratory, Newport, RI 02841) NUSC IR/IED. ]
TUESDAY AFTERNOON, 17 MAY 1988 WEST BALLROOM A, 1:45 TO 4:50 P.M.
SessionP. PhysicalAcoustics II: Martin GreenspanMemorial Session
Steven L. Garrett, Chairman Departmentof Physics,Code 61GX, Naval Postgraduate School Monterey, California 93943
Chairman's Introduction--l:45
Invited Papers
1:55
PI. Martin Greenspan'swork in aeoustiealphysics. Richard K. Cook(4111 Bel Pre Road,Rockville, MD 20853) Martin(Moe) Greenspan's long career at theNational Bureau of Standards began in 1936in thegeneral areaof thestrength of engineeringmaterials and structures. His theory of stressdistribution in perforated plateswas an important contribution to modernlinear elastic fracture mechanics. In 1947he began to workin physicalacoustics and continued in thisuntil 1987.His landmarkaccomplishments include the theoretical analysisand measurement of sound propagation in rarefiedgases, development of aninstrument for measure- mentof sound velocity in seawater, and determination ofthe important physical influences leading to cavita- tionin liquidsby ultrasound.In recentyears, he worked on methods for measuringaccurately the acoustical emissions(stress waves) generated by physically stressed materials of engineering.The influences of hiswork on soundin rarefiedgases persist today; the database he obtained in the late 1940s-1950swas far in advanceof the analyticalcapabilities for soundpropagation in rarefiedgases at that time.The dataare still usedfor evaluationof newtheoretical developments. Moe's work has always been characterized by an excellent coordi- nationbetween theoretical-mathematical development and laboratory experiment.
2:20 P2.Acoustic goofs '• or irreproducibleeffects in acoustics.I. Rudnick (Physics Department, UCLA, Los Angeles,CA 90024) Ourrespected and beloved friend, Moe, had a geniusfor being able to puthis finger on mistakes that made theirway into experiments and theories. He wasa bubblebuster. Listed below are some acoustic goofs. Moe contributedhis skill in correctingsome of them.There will bebrief discussions of some of thetopics as time allows:( 1) Fitzgcrald'sfolly; (2) The universalsequence in subharmonicgeneration or whenis a decibelnot a decibel?;( 3 ) Whatis zero sound in liquidhelium 4?; (4) Isothermaland adiabatic propagation in metals;( 5 ) Shearingin longitudinalwave propagation; (6) Weber'sfolly. a• Webstcr's dictionary: goof--to make a mis- take or blunder.
2:45
P3. The dispersionof soundin monoatomicgases. George W. Ford (Departmentof Physics,University of Michigan, Ann Arbor, MI 48109-1120) The classicalKirchhoffderivation of theabsorption of soundin gasesat moderatepressures starts from the Navier-Stokesequations of hydrodynamics.At low pressures, specifically when the mean free path is no longer
$38 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115thMeeting: Acoustical Society of America S38 smallcompared with the soundwavelength, this same derivation predicts a dispersionof the speedof sound. The absorptionand dispersioncan also be deriveddirectly from the Boltzmannequation, with a resultthat confirmsthe Kirchhoffabsorption coefficient, but with predicteddispersion and higher-orderterms in the absorptionthat differfrom thoseof hydrodynamics.Martin C•reenspan'searly work on soundpropagation at lowpressures provided the experimental data for a definitivetest of theBoltzmann equation. These derivations will bebriefly reviewed and the comparison with Crreenspan'sresults will beshown. Some concluding remarks aboutthe historyof this problemand aboutmore recent developments will be made.
3:10
P4. Adiabaticinvarlance, the cornerstoneof modernphysics. Seth J. Putterman(Department of Physics, UCLA, Los Angeles,CA 90024) The articleby M. Greenspanon the Boltzmann-Ehrenfestadiabatic principle [J. Acoust.SOc. Am. 27, 34 ( 1955) ] isused as the starting point for discussingthe role of adiabaticinvariance (or integrability) in modern physics.These conserved properties of opensystems constitute the keyconcept upon which all advancesin modernphysics have been focused. These include the origin of thermodynamicirreversibility for Hamiltonian systemsand the quantizationof motion.Delauney (1860) wasfirst to employadiabatic invariants as dynami- calvariables with an aimtoward determining the stability of thesolar system. This issue eventually culminated in the work by Kolmogorov,Arnold, and Moseron the stabilityof nearlyintegrable systems. The solutionto the problemof solironmotion by the inversescattering transformation is in fact a transformationto the adiabaticinvariant as the canonical momentum. Recent acoustical applications of adiabaticinvariance include a determinationof the potentialin an acoustic levitator and a predictionof second sound in acousticturbulence.
3:35
PS. Gas-filledspherical resonators: The currentstate of experimentand theory.James B. Mehl (Physics Department,University of Delaware,Newark, DE 19716)and MichaelR. Moldover(Thermophysics Division,Center for ChemicalEngineering, National Bureau of Standards,Gaithersburg, MD 20899} In thepast decade, spherical acoustic resonators have proved to be useful tools for high-accuracy measure- mentsof thespeed of soundin gases.Applications include thermophysical property determinations, acoustic thermometry,and the recent redetermination of the gas constant. The complex resonance frequencies of the resonatorare described by a theoreticalmodel that includes the effects of theviscous and thermal boundary layers,shell motion, imperfect spherical geometry, and other, less important, effects. The theoretical model is testedby comparingthe consistency of speed-of-sound determinations using different modes, and by compar- ingthe resonance half-widths with thetheoretical predictions. The currentlevel of agreementof experiment and theorywill be reviewed.
P6. Cavitation nuclei and cavitation noise.M. Strasberg(David Taylor ResearchCenter, Bethesda,MD 20084-5000)
Presentunderstanding of the influenceof nucleion the onsetof cavitationand the generationof radiated cavitationnoise is reviewed.It is now generallyaccepted that the formationof cavitiesin a liquidat only moderatenegative pressures requires the presenceof nucleiin the liquid;the experiments of Greenspanand Tschiegg,among others, leave no doubt about this. However, there is still no agreement concerning the form of thesenuclei or the mechanismcausing them to persistunder conditions that shouldordinarily make them disappear.Various possible stabilizing mechanisms are discnsscd. It is shownthat althoughnuclei determine whethercavitation occurs at all, onceit occurs,their presencehas little effecton the characteristicsof the radiatednoise generated by thecavitation. The factors that do influence the spectrum and the magnitude of the noisewill be discussed,mainly in termsof hydraulicallyinduced cavitation.
4:25
P7.Theory of radiatianinduced cavitation. Robert E. Apfeland Y.-C. Lo (Yale University,P.O. Box2159, New Haven, CT 06520) Datafrom two different experiments fall on the same general nondimensional curve: The first set of datais that of Greenspanand Tschieggon neutron-inducedacoustic cavi•ation [J. Acoust.Soc. Am. 72, 1327 (1982) ]; thesecond is thatof Apfel et aL on neutron-induced(nonacoustic) cavitation of superheateddrops [Phys.Rev. A 31, 3194 (1985)]. The physicsis the same.The theorybegins with the neutron-nucleus interaction,continues as the recoil ion deposits its energy as heat along a line,and ends with theformation of a bubblethat must exceed a criticalsize ifa macroscopicbubble is to beproduced. The theorywill be reviewed andits predictions compared with both sets of experimentalresults. [Work supported by D. O. E. grantDE- FG02-87ER6050. ]
S39 J.Acoust. Soc. Am. Suppl. 1, Vol. 83, Spring 1988 115thMeeting: Acoustical Society of America S39 WEDNESDAY MORNING, 18 MAY 1988 ASPEN ROOM, 8:00 A.M. TO 12:30P.M.
SessionQ. Educationin AcousticsII: In-situ TechnicalTour of NorthwestAcoustical Organizations
Gerald F. Denny, Chairman HoneywellMarine SystemsDivision, 6500 Harbour Heights Parkway, Everett, Washington 98204
VariousNorthwest organizations will havedisplays and representatives available from 8:00a.m. to 12:30p.m. to acquaintvisitors with theirfacilities and their workin acoustics.
Q1. AppliedPhysics Laboratory (University of Washington,1013 N.E. 40th Street,Seattle WA 98105). The laboratoryfocuses on education, research, and applied technology in ocean,polar, and environmental sciences.
Q2. Boeing(Seattle, WA 98124). The BoeingCommercial Airplane Noise Technology Staff is involvedin the designand development of noisesuppression hardware and noiseprediction and testing technologies.
Q3. David Taylor ResearchCenter, DetachmentPuget Sound(Bremerton, WA 98314). The detachment conductsresearch, development, testing, and evaluation in underwateracoustics, and submarine noise reduc- tion.
Q4. Honeywell,Marine SystemsDivision (6500 Harbour HeightsParkway, Everett,WA 98204). MSD developsand produces underwater transducers, sonar systems, and acoustic signal processing equipment.
Q$. Naval UnderseaWarfare EngineeringStation (Keyport, WA 98345). NUWES is activein researchand measurementof underwateracoustic phenomenon.
Q6. Universityof Washington(Seattle, WA 98195).Departments that teach and carry out •search in acous- ticsinclude Aeronautics and Astronautics, Biocngineering, Electrical Engineering, Mechanical Engineering, Music,Otolaryngology, Physics, and Speech and HearingSciences.
Q7. VeteransAdministration Medical Center ( 1660S. ColumbiaWay, SeattleWA 98108). The Audiology Clinicof theVA is carryingout cochlearimplants as part of a researchproject for veteranswith severehearing losses.A demonstrationof the signalprocessing strategies of threecochlear implants used in their eochlear implantproject will begiven.
WEDNESDAY MORNING, 18 MAY 1988 EAST BALLROOM B, 8:00 TO 11:20 A.M.
SessionR. EngineeringAcoustics IIl and Noise IV: Measurementof PressureFluctuations in Turbulent Boundary Layers
Allan Zuckerwar, Chairman N/ISm4Langley Research Center, M/S 238, Hampton, Virginia23665
Chairman's Introductions8:00
Invited Papers
8.'O5
RI. Acousticsources in the turbulentboundary layer. I. C. Hardin (NASA LangleyResearch Center, MS- 461, Hampton,VA 23665) This paperwill beconcerned with sourcesof soundin thesimplest case of a turbulentboundary layer over a flat plate. For the majorityof the paper,the drivingflow will be consideredto be subsonicsuch that gross compressibilityeffects, such as shocks, do notoccur. Standard terminology, governing equations, and charac- teristicsof theturbulence in theboundary layer and the noise it produceswill bedescribed. The paperwill then examinethe fundamentalinstability of this flow at high Reynoldsnumber and the transitionof a laminar boundarylayer to turbulence.Generation of Tollmien--Schlich ting waves and the myriad large scale structures (hairpinvortices, streaks, spots, bursts, etc. ) thathave been observed in boundarylayers will alsobe described.
,•10 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America •40 Finally,the theoretical understanding ofnoise production bythis flow, which was initiated by Lighthill, Curie, andPowell and carried on by laterresearch workers, will bedescribed. In particular,the useof different Green'sfunctions, which allows the noise to be determined either through integration ofvarious flow quantities overthe volume of theboundary-layer flow or by integrationof thepressure over the flat plate, will be developedand the advantages of each noted. In lightof thisunderstanding, mechanisms forboundary-layer noiseproduction and methods for their calculation are discussed. Additiolaal sources in high-speed boundary layerswith shocks present will alsobe described.
8:35
R2. A semiempiricalmodel for the wavevector-frequency spectrum of turbulentwall pressureon a smooth planarboundary. D. M. Chase(Chase, Inc., Suite 510, 87 SummerStreet, Boston, MA 02110) The ana•tical frameworkand construction of a modelof turbulentboundary-layer pressure on a planar wallby its relationto velocity-productfluctuations regarded as its .sources are reviewed. Expansion for low (subconvective)wavenumbers and low Mach numbersprovides a basisfor sucha trial modelof the wall- pressurespectrum that is potentiallysatisfactory at all wavenumbersfrom theacoustic domain to the convec- tive.The proposedform for sourcecross spectra incorporates the principleof wall similarity,a kinematic assumptionabout space-time correlation, possible wall-normal profiles of vorticitycross spectra, and assumed non-negativity.The wall-pressure model is obtained by analytical integration over the source-spectral profiles in limitingdomains followed by convenient rough interpolation. The experimental basis for determination of parametervalues of themodel is considered by referenceto classicalwind-tunnel measurements dominated by the convectivewavenumber domain and to moreambiguous ones directed to the subconvectivebut super- acousticdomain. The inadequatestate of validationand determination of parameterscharacterizing the acous- ticdomain and the domain of pertinenceof theKraichnan-Phillips theorem is confronted. Likewise, the open but experimentallyresolvable question of thejoint dependenceon wavevector components in the convective domain is recalled.
9:05
R3. The influenceof surfaceroughness on wall-pressurefluctuations and the radiated soundfrom a turbulent boundarylayer. M. S. Howe (BBN Laboratoriesinc., 10 Mou]tonStreet, Cambridge, MA 02238) The noiseand vibrationproduced during turbulent boundary-layer flow overa nominallyplane flexible surfaceare governed by the low wavenumber and acoustic regions of thewall-pressure wavenumber-frequency spectrum.In thecase of a roughsurface, the spectrum differs from that on a smoothwall on two counts:(i) the strengthsof theturbulent Reynolds stress fluctuations, which are ultimately responsible for the pressure field, areincreased by the actionof surfaceroughness; (ii) the pressurefield produced by thoseenhanced pressure sourcesis redistributedin thewavenumber plane by diffractionby the roughnesselements. The wall-pressure spectrumcan be expressedin the form P(k,•o) = Po(k,•o)+ PR(k,o), wherePo denotes the spectrumthat wouldbe associatedwith the roughness-enhancedReynolds stresses if the wall wereassumed to be perfectly flat, and PR is the additionalcomponent due to the diffractionmechanism OiL It will be shownthat PR is expectedto dominatethe behaviorof thewall pressurein theIow-wavenumber and acoustic regions over a wide rangeof frequencies.Certain problems associated with experimentalstudies of roughwall boundary-layer noise will also be discussed.
9:35
R4. Use of piezoelectricfoil for measurementof pressurefluctuations and local shear stressin flight. A. Bertelrud(High TechnologyCorporation, Hampton, VA 23366and FFA, The AeronauticalResearch Institute of Sweden, Bromma, Sweden)
Themeasurement of pressurefluctuations in air flowsposes several serious problems to theexperimentalist. Generallya very limited frequencyresponse is available,as the cavityabove ordinary pressure transducers playsa crucialrole for the sensorresponse. This type of transducerrequires extensive work for incorporation into a surface.Recently, piezoelectric foil hasbecome available for variousapplications. It offersease of applicationand distributedmeasurements without strict frequencyresponse limits. However,although the dataacquisition is straightforward, it appears that particular caution is required for interpretationof thedata as pressurefluctuations. In the presentpaper, emphasis is put on frequencyresponse, true root-mean-square determination,as well as interpretation into skin friction data. Through analysis of flightdata and laboratory calibrations,it isshown that the piezoelectricfoil in generalwill respondto pressurefluctuations, and it isalso shownthat the piezofoilconstitutes a repeatable and rugged source of information.
10:0.5-10:.20
Break
$41 J. Acoust.Soc. Am. Suppl. 1. VoL 83. Spring1988 115th Meeting:Acoustical Society of America S41 ContributedPapers
10:.20 , 10:50
RS. Bubbleimth in the wake of a eavitntingpropeller. Manro Pierueci R?. A ealnbratorfor the measurementof the static and dynamicsbe•r ( Departmentof AcrospaeeEngineering and Engineering Mechanics, San stres•ea along a stretched Mylar memlirene. Frank W. Cuomo DiegoState University, San Diego, CA 92182) (Departmentof Physics,The Universityof RhodeIsland, Kingston, RI 02881-0817and Naval UnderwaterSystems Center, Newport, R102841- Propellersthat operateunderwater at high rpm'scavitate at the tip. 5047) The tip cavitationcreates air bubblesthat arethen swept downstream by themotion of thesurrounding fluid. In thispaper, a theoryis presentedto Recentpower spectral density measurements of the pressurefluctu- predictthe local velocity and the path of thebubble. The bubble motion is ationsin a turbulentboundary layer obtainedwith a fiber-opticlever assumedto begnverned by a groupof termsdue to theacceleration of the pressuresensor in a low-speedwind tunnel have compared favorably with displacedfluid, the convective term, and the drag due to thecruss-section- data reportedin the literature.Utilizing a similarsensor arrangement, al areaof the bubble.At verylow and veryhigh Reynolds numbers, the incorporatinga flush-monnted,metallized Mylar membrane,the mea- equationshave been solved in closedform. Results are presented for the surementof shearstresses leading to skinfriction has been contemplated. bubblevelocity and pathfor the followingflow fields:(a} uniformaxial The firststep toward the realization of a shearsensor has been to establish flow field and (b) uniformflow field with an axially decayingswirling thebehavior of a stretchedMylar membraneacted upon by forcesparallel component.In all easespresented the bubbleaxial velocitycomponent to the flow direction.A calibratordesigned to meet the requirements asymptotesto thefree stream velocity; the manner in whichit asymptotes normallyfound in a wind-tunnelenvironment has been constructed and is exponentialat verylow Reynoldsnumbers and algebraic at highReyn- tested.In thispaper, its operation is described and data of Mylar samples oldsnumbers. Bubble helical paths and velocitypatterns are shownfor are presented.The resolutionof the devicefor staticmeasurements is different bubble sizes. basedon translationson the orderof I/.tin. whilethe dynamicresponse coversa frequencyrange ofdc to 18 kHz.
10:35 11:05
R6. Extendedfrequency response of n fiber-opticmicrophone. Allan R8. Comparisonbetween design and installedacoustic characteristics of J. Zuckerwar(NASA LangleyResearch Center, MS-238, Hampton, VA the NASA Lewis 9-X15-ft low-speedwind-tunnel acoustic treatment. 23665) and Mathew L. Areford (Hampden-SydneyCollege, Hampden- Milo D. Dahl and Richard P. Woodward (National Aeronautics and Sydney,VA 23943) SpaceAdministration, Lewis Research Center, Cleveland, OH 44135) A condensermicrophone responds to the meandisplacement of a The testsection of theNASA Lewis9-X 15-ftlow-speed wind tunnel stretchedmembrane excited into an axiallysymmetric mode of vibration. wasacoustically treated to allow the measurementof soundunder simu- At somefrequency between the first and second resonant frequencies, the lated free-fieldconditions. The treatmentwas designedfor high sound mean displacementfalls to zero and the microphoneresponse drops absorptionat frequenciesabove 250 Hz andto withstandthe environmen- sharplyafter the firstresonance. A fiber-opticmicrophone, on the other tal conditionsthat existin the testsection. To achievethe design require- hand,responds to the displacementat the centerof the membrane.The meats,a fibrousbulk absorbingmaterial was packed into removablepanel centerdisplacement of an axiallysymmetric mode never falls to zero. sections.Each section was divided into two equal-depthlayers that were Consequently,through judicious backplate design, the membrane motion packedwith materialto differentbulk densities.The lower densitywas canbe damped to achievea smoothmerger in responsebetween the first nextto theface of thetreatment. The facingconsisted of a perforatedplate and secondresonances, and the bandwidthof the microphonecan be and screeningmaterial layeredtogether. Sample tests for normal inci- extendedto beyondthe secondresonant frequency. After a briefreview of denceacoustic absorption were conductedin a small rectangularduct. theconstruction of thecondenser and fiber-optic microphones, the results Testswith no air flow involvingthe measurementof the absorptiveprop- of a theoreticalanalysis will bepresented for twocases: ( 1 ) a B&K t-in. ertiesof theinstalled treatment combined the use of timedelay spectrom- condensermicrophone compared with its fiber-opticequivalent, and (2) etry with a previouslyestablished free field measurementmethod. These •a-in.fiber-optic microphone of specialdesign to measurepressure fluctu- measurementswere compared to thedesign model and sample tests of the ations in a turbulent boundary layer. treatment.
•42 J.Acoust. Soc. Am. Suppl. 1. Vol. 83. Spring 1988 115thMeeting: Acoustical Society of Amedca S42 WEDNESDAY MORNING, 18 MAY 1988 GRAND BALLROOM A, 8:10 A.M. TO 12:00 NOON
SessionS. Physical AcousticsIII: Lord Rayleigh's Contributionsto Acoustics
HonorableGuy R. Strutt, Cochairman OM Rectory,Terling, Chelmsford, Essex CM3 2QE, England
Malcolm J. Crocker, Cochairman Departmentof MechanicalEngineering, Auburn University,Auburn, Alabama 36849
Chairman's Introduction--8:10
Invited Papers
8:15
Sl. Lord Rayleighand his influenceon scienceand scientists.Malcolm J. Crocker (Departmentof MechanicalEngineering, Auburn University, Auburn, AL 36849) JohnWilliam Strutt,the third Lord Rayleigh( 1842-1919),was a manof considerablescientific accom- plishments.He contributedto theoreticaland experimental physics and was one of the greatpioneers in scientificand engineering methods. He wasalso a manwho moved in thehighest social and intellectual circles of hisday. He wasa relativeof primeministers and received almost every honor to whicha Britishscientist couldaspire. He wasa Nobelprize winner and President of theRoyal Society. His book, The Theory of Sound, hasremained in print for overa century.From 1880to 1885,after J. C. Maxwell'sdeath, Lord Rayleigh directedthe famedCarendish Laboratory at CambridgeUniversity, but after that mainlylived in Terling Place,his country house near London where he established a remarkable scientific laboratory in whichhe later discoveredargon. During his lifetime Lord Rayleigh published over 400 scientificpapers and carried on a vigorouscorrespondence with scientists such as Maxwell, Kelvin, Stokes and others. This paper will examine hispapers and correspondence to put hisscientific contributions, particularly in acoustics,into historical perspective.
8:45
S2. Rayleigh'sphilosophy of problemsolving, as inferredfrom his original paperson acoustics.Allan D. Pierce(School of MechanicalEngineering, Georgia Institute of Technology,Atlanta, GA 30332) Rayleighnot only had a knackfor skimming otf the cream, but also for findingcream where others might havethought there was no cream to be found. His firstsubstantial work of note,the 1870memoir on resonance in thePhilosophical Transactions, introduced a styleand philosophy that is evident in manyof hislater works and that is worthemulating. In thispaper, Rayleigh successfully bypassed the elegantbut difficultto apply mathematicsof Helmholtz'sepochal 1860 paper, and substituted instead a few simpleapproximate concepts andproduced some relatively simple formulas that could be compared with Sondhauss's experimental results onwhat we now call Helmholtz resonators. Other examples cited in thepresent paper are Rayleigh's theory of diffractionby orificesand slits, his analytical models for elasticshells, his discovery of adiabaticinvariants in his 1902"pressure of vibrations"paper, and his approach to modelinglow-frequency sound radiation from sources.Rayleigh's work and philosophymay havehad a substantialinfluence, although perhaps not con- sciouslyrecognized, on modern trends such as finite elements and matched asymptotic expansions. In addition to a thoroughfamiliarity with basicphysical concepts, a goodintuitive feel for what wasrelevant, and an inherentlove of simplicity,Rayleigh made extensive use of variationaltechniques as a frameworkfor develop- ingapproximations. Hitherto they had been used primarily as a meansfor elegantexpression of basicprinci- ples.Also noteworthy was Rayleigh's heavy use of similitudearguments. [Work supportedby ONR.]
9:15
S3. Rayleigh'sinfluence on engineeringvibration theory. Stephen H. Crandall(Massachusetts Institute of Technology,Cambridge, MA 02139) Thefirst ten chapters of TheTheory o/Sound constitute an unprecedented exposition of vibrationtheory. This tour de force setthe mold for the teachingof vibration for more than a century.The central core is linear modalanalysis applied to lumped parameter systems and to strings, bars, beams, membranes, plates, and shells. In addition,there are preliminary investigations ofnonlinear vibrations, parametrically excited vibrations, and randomvibrations. Most of the textbookson vibrationsfor engineersthat haveappeared in the past50 years maybe viewed as attempts to explainRayleigh's idea to universitystudents. Rayleigh's paradigm has served us
S43 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115th Meeting:Acoustical Society of America S43 wellfor a centurybut it appearsthat we are now on the threshold ofmajor changes in the way vibration theory istaught. These changes are being driven by the tremendous advances ininstrumentation anddata processing facilities.The second coming of modalanalysis (modal testing, really) is largelybased on the concept of transferfunctions. Although Rayleigh occasionally used the complex exponential and even computed a few transferfunctions in passing,he did not attach any special significance thereto. It isinteresting tospeculate whatRayleigh would have done had he envisioned today's multichannel spectral analyzers. Other advances, unavailableto Rayleigh,which are currently entering vibration courses, are discussed.
9:45 S4.Lord Rayleigh and reciprocity in physics. Richard K. Cook(4111 Bel Pre Road, Rockville, MD 20853) In TheTheory ofSound, Rayleigh presents hisanalysis ofthe reciprocal theorem ofmathematical physics with the observationthat "Very remarkablereciprocal relations exist between the forcesand motions of differenttypes...." ina linearized dynamical system. He introduces thereciprocity principle for sound waves in anair-filled space: "On his extension ofGreen's theorem, Helmholtz founds his proof of the important (reci- procal)theorem contained in the following statement: ...."In theensuing 100 years since publication ofhis book,the reciprocity principle has been extended tomany areas of physics---electrodynamics, electroacoustics, piezoelectricity,electromagnetic waves, optics, etc. The principle has been successfully applied inelectroacous- ticsto achievea system for accurate, absolute calibration of microphonesused for acoustical measurements. Thesystem isnow internationally usedby standards laboratories. Limitations and restrictions onapplicability ofthe reciprocity principle tocertain physical systems have come to light. Conspicuous examples involve the propagationofsound in moving air and the propagation ofelectromagnetic waves through magnetized media.
10:15
S5. Lord Rayleigh'scontributions to musicalacoustics. Thomas D. Rossirig(Department of Physics, NorthernIllinois University,DeKalb, IL 60115) Rayleigh'swork touched virtually every area of acousticsin someway• musical acoustics iscertainly no exception.Besides laying the groundwork for all subsequentresearch on musical instruments with his informa- tive discussionsof the vibrationsof strings,bars, membranes, plates, air columns,and resonators,he did importanttheoretical and experimentalwork on bellsand kettledrums.He identifiedthe vibrationalmodes thatradiated the most prominent partials in thesounds of boththese instruments, and he recognized that the nominalpitch of a churchbell lies one octave below the fifth partial. In thearea of musicperception, his most importantcontributions were his studies of absolutepitch and sound source localization.
10:4• S6.Lord Rayleigh's contributions to aeroaconstics. Alan Powell(Department of MechanicalEngineering, Universityof Houston,Houston, TX 77004) Thegrowth of aeroacoustics asa distinct discipline burgconed at midcentury, initially fostered by the need tocontrol jet noise. It mergedthe then separate disciplines ofaerodynamics and sound. Rayleigh knew no such boundariesand would likely have made major contributions toaeroacoustics if the fundamental questions had capturedhis attention. As it was,he wrote on many highly pertinent aspects: e.g., his theory for sensitive flames andjets (visual sound indicators) led to the modern instability theory of jets and other important aerodynamic flows,while the hypothesizedfeedback in his"bird-call" (a compatiblehigh-frequency sound source) is similarto thatin edgetones and in manyother flow resonances. These had puzzling inconsistencies until Rayleigh'smethod of similitude was applied through Reynolds and Strouhal numbers. He hadapplied it to Strouhal'saeolian tones and to thescattering of soundfrom inhomogeneities, in which he introducedthe "acousticmodel" of aerodynamic sound theory (and in whichthe method for Rayleigh's integral was used for planeboundaries). There are now hundreds of researchpublications in aeroacoustics, some notable books and a placein universityeducation: Rayleigh's name occurs many times.
11:15 S7.Lord Rayleigh's contributions to nonlinear acoustics. Robert T. Beyer(Department of Physics,Brown University,Providence, RI 02912) Whilethe overwhelming majority of Rayleigh's contributions to acoustics have been in thelinear domain, thereare many nonlinear topics on which he made early progress and set the stage for further progress in the future.His work on radiation pressure, cavitation, shock waves, and flow-induced sound will bediscussed and the developmentsthat followedfrom them will be described.
S44 J.Acoust. Soc. Am. Suppl. 1, Vol. 83, Spring 1988 115thMeeting: Acoustical Society of America S44 ContributedPaper
11:45 When the organkeys for d'" (1174 Hz) and e'" (1318 Hz) are played simultaneously,a gravenote of 144 Hz (betweena C# and D) is also S8. Rayleigh'sobservations of waveinterference, evanescent fields, and heard,giving a displeasingsound. Rayleigh carefully tuned the pipesso soundcancellation. David C. Swanson(Textron DefenseSystems, that the gravenote gave the slowest possible beat with a tuningfork of 128 Acoustic/SeismicSensor Technology Group, 2385 Revere Beach Hz. Rayleighnoted that if the gravenote was due to the ear, the soundof Parkway, Everett, MA 02149) the tuningfork wouldnot cause the slow cycles of soundcancellation. He alsonoted that the soundfield from the fork-pipecombination (during In Chap.23 of TheTheory of Sound, Rayleigh made some outstanding constructiveinterference) fell off more rapidlywith distancethan the observationsof waveinterference in an experimentdesigned to showthat sound field from the fork alone, which is characteristicof an evanescent- the sourceof "gravenotes" in pipe organswas due to differencetones or nearfield.These observations provide an interestinglesson for all ex- ratherthan a nonlineafityin the earas proposed by Helmholtz[J. W. S. perimentalistson basicscientific reasoning and the importanceof report- Rayleigh,The Theoryof Sound (Dover, New York, 1945), pp.461-462 ]. ing all relevantobservations revealed by experiment.
WEDNESDAY MORNING, 18 MAY 1988 GRAND BALLROOM C, 8:30 TO 9:50 A.M.
SessionT. SpeechCommunication III: Neural Networksin SpeechRecognition I
Les Atlas, Chairman Departmentof ElectricalEngineering, FT-10, Universityof Washington,Seattle, Washington 98195
Chairman's Introduction4:30
Invited Papers
8:35
T1. Neuralnets for speechrecognition. William Y. Huang,Richard P. Lippmann,and Thao Ngnycn(P.O. Box ?3, MIT Lincoln Laboratory,Lexington, MA 02173) Artificial neuralnetworks arc of interestfor two main reasons.First, they providearchitectures to imp]e- mcntmany algorithms used in speechrccognizcrs with finegrain massive parallelism. Second, they are leading to newcomputational algorithms and newapproaches to speechrecognition inspired by biologicalnervous systems.Neural net approachesto the problemsof speechpreprocessing, pattern classification, and time alignmentare reviewed. Preprocessors using auditory nerve time-synchrony models have provided improved recognitionperformance in noise[O. Ghitza,ICASSP 87, 2372-2375].Highly parallel neural net architec- turesexist to implementmany important traditional classification algorithms, such as k-nearest neighbor and Gaussianclassifiers [R. Lippmann,IEEE ASSP Mag. 4(2), 4-22 (1987)]. Newer multilayerperceptron classifierstrained with backpropagation can form arbitrary decision regions, are robust, and train rapidly for convexdecision regions. These nets performed as well asconventional classification algorithms for digitand vowelclassification tasks [W. Huanget al., 1stICNN IV, 485 (1987) ]. Neuralnet approaches to theproblem of timealignment are reviewed, including a neuralnet modelcalled a Viterbinet that implementsthe Viterbi decoderused successfully in continuousdistribution hidden Markov models (HMMs) [R. Lippmannet al., 1stICNN 1¾,417 (1987) ]. [This work wassponsored by the DefenseAdvanced Research Projects Agency and the Departmentof the Air Force.]
9:00
T2. Speechrecognition using time-delay neural networks. A. Waibel,T. Hanazawa,K. Shikano(ATR InterpretingTelephonyResearchLaboratories, Twin 21 Tower, Osaka 540, Japan),G. Hinton (University of Toronto,Toronto, Canada), and K. Lang (Carnegie-MellonUniversity, Pittsburgh, PA 15213) A time-delayneural network (TDNN) approachis presented to speechrecognition that is characterized by twoimportant properties: ( 1 ) Usingmultilayer arrangements of simple computing units, a TDNN canrepre- sentarbitrary nonlinear classification decision surfaces that are learnedautomatically using error back propa- gation.(2) The time-delayarrangement enables the networkto discoveracoustic-phonetic features and the temporalrelationships between them independent of positionin timeand, hence, not blurred by temporal shifts in the input.The TDNNs are comparedwith the currentlymost popular technique in speechrecognition, hiddenMarkov models (HMM). Extensiveperformance evaluation shows that the TDNN recognizesvoiced stopsextracted from varying phonetic contexts at anerror rate four times lower ( 1.5%vs 6.3% ) thanthe best of ourHMMs. To performthis task, the TDNN "invented"well-known acoustic-phonetic features (e.g.,F2
$45 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115thMeeting: Acoustical Society of America S45 rise,F2 fall, vowelonset) as usefulabstractions. It alsodeveloped alternate internal representations to link differentacoustic realizations to the sameconcept. The TDNNs trainedfor other phoneticclasses achieve similarhigh levelsof performance.The integrationof suchsmaller networks into largephonetic nets and proposestrategies for the designof neuralnetwork based large vocabulary speech recognition systems is discussed.
9:25
T3. Speechrecognition with layerednetworks. D. J. Burr (Bell CommunicationsResearch, Morristown, NJ 07960)
Experimentsshow that pereeptronsmay be usefulfor speechrecognition. Pereeptrons, or layerednet- works,can recognizespoken digits, polysyllabic words, and alphabetictext with accuraciesexceeding 98%. They do so with variedinput representationsranging from Fourier throughcepstral and linear prediction coefficients.As in conventionalclassifiers, the greater the numberof trainingtokens per class, the higher is the accuracy.Trained networks may be analyzed to observeany structure in theweight space. One can see evidence in perceptronnetworks for both•oicing andformant onset detection capability. For a givenproblem, a layered networkhas a criticalnumberof hiddenunits that maximizesrecognition accuracy at minimumhardware cost. Additional hardwaremay be savedby eliminatingnear-zero weights in upper layersas thesehave minimal effecton recognitionaccuracy. A tradeoffmay existbetween number of weightsand networkcomplexity. A small layerednetwork with 264 weightsand 12 hiddenunits achieved99% accuracyon isolateddigits. In contrast,a largepereeptron with morethan 11000 weightsand no hidden units achieved 98.2% accuracyon thedifficult E-set alphabet and perfect recognition on twentypolysyllabic words. Apparently, memory (num- berof weights)can be tradedfor logic(number of hiddenneurons} as in conventionalcomputer programs.
WEDNESDAY MORNING, 18 MAY 1988 EAST BALLROOM A, 8:30 A.M. TO 12:05 P.M.
SessionU. UnderwaterAcoustics III: StochasticVolume and BoundaryScattering III
Erie I. Thorsos, Chairman .4ppliedPhysics Laboratory, Unioersity of Washington,1013 N.E. 40th Street,Seattle, Washington 98105
Chairman's Introductions&30
ContributedPapers
8:35 8:S0
U1. Reflections from a block of sea ice. Gerald R. Garrison and Robert U2. Reflectivityof sealee. RobertE. Francois,Gerald R. Garrison,and E. Francois(Applied PhysicsLaboratory, University of Washington, Timothy Wen (AppliedPhysics Laboratory, University of Washington, 1013 NE 40 Street, Seattle, WA 98105) 1013 NE 40 Street, Seattle, WA 98105)
At an Arctic ice campin the BeaufortSea, monostafic acoustic reflec- The reflectionsfrom the face of a finite block of sea ice are usually tions near normal incidence were measured from the lower face of an predictedas the combinationof two phenomena,as allowedunder the isolatedblock of natural seaice. The measurementswere repeatedwith Kirehhoffapproximation. One is the reflectionloss due to thechange in the transitionlayer sawedoff to leavea surfaceof solid ice. A comparison acousticimpedance at an infinitewater-ice boundary. The other is the of the two resultsshowed that the 15-era-thickskeletal layer and transi- interferencepattern produced by the coherentreturn from a finiteblock tionzone cause a reductionin reflectivity,beyond that predictedfrom the face. Measurements of the reflections from a block of natural sea ice, impedancemismatch, that increasesfrom 10db at 20 kHz to 30 dB at 80 describedin a companionpaper, showed that the reflectionis further kHz. The systemwas calibrated using the air-water interfaceformed at reducedby the skeletal layer. More information on the acoustic properties the openbottom of a shallowhollow metal cylinder. Air wasforced into of this layer is neededin order to modelits effect.Reflection losses are thecylinder when it wasin placeto fill it andthus provided a flat pressure- describedby an amplitudereflection coefficient which would be 0.34 due releasereflecting surface. As the measuringtransducer was movedaway to theimpedance mismatch at thewater-ice boundary. Reflection coeffi- from normalincidence, the responsepattern was found to agreewith the cientsmeasured for 1-to 2-m-thickice in theArctic at frequencies20-80 sidelobepattern predictedfor a flat, perfectlyreflecting surface. These kHz arecompared with measurements byStanton and Jezek for thinice. measurementsof the effectof the skeletallayer on ice reflectivity will be Duringthe Arctic measurements, the returnfrom the back face of the helpfulin modelingthe reflectionsfrom icekeels. [ Work supportedby the blockwas also detected, providing information on the absorption of sound Offce of Naval Technologywith technicalmanagement by NORDA. ] in theice. Field experiments have been designed to separate the absorption
S46 J. Acoust.Sec. Am. Suppl.1, Vol. 83, Spdng1988 115th Meeting:Acoustical Society of America $46 in theskeletal and transition layer from the absorption in solidice. [Work proacheszero to reproduceone-component results); more generally, the supportedby theOffice of NavyTechnology with technicalmanagement curveis skewed, and as v approachesunity,œreduces to the exponential for by NORDA. ] the simplestPoisson case. Numerical integrations over finite ranges of d provideillustrative graphs to exhibiteffects of truncationand parameters on the reflectioncoefficients and differentialscattering cross sections per unit area.[Work supprrtedby ONR.] a•Visiting from Department of 9:05 MathematicalSciences, Loyola University, Chicago, IL.
U3. New findings on scattering from floating plate with lateral heterogeneities. Jacques R. Chamuel (Sonoquest/Advanced Ultrasonics Research,P.O. Box 153, Wellesley Hills, MA 02181 ) and 9:50 Gary H. Brooke (Defence Research EstablishmentPacific, FMO Victoria, British Columbia V0S lB0, Canada) U6. Application of Twersky's boss scattering theory to laboratory measurementsof sound scattered by a rough surface. D. Chu and T. Experimentalultrasonic modeling results are presentedon neargraz- K. Stanton( 1215West Dayton Street, Geophysical and Polar Research ing incidencescattering from cracks, ridges, and edges of floatingplates Center,University of Wisconsin,Madison, WI 53706) boundinga shallowwater waveguide.It wasobserved that underwater acousticwaves propagating in a shallowwater waveguide bounded by a Oneof the manychallenges in describingthe scatteringof soundby floatingplate produce a ringingwave component in the waterupon scat- roughsurfaces is to addressthe fact that most surfaces are three dimen- teringfrom a crackor a plateedge. This ringing component isassociated sional. Furthermore,only their statisticalproperties may be known. with the waterwave, travels with the waterwave velocity, and its time of Twersky'smodel IV. Twersky,J. Acoust.Soc. Am. 29,209-225 (1957) ], arrivalis independentof thelength of plateregion along the wave path. by usingan imagemethod and takingmultiple scattering into account, Theringing frequency, however, is determined by the shear wave velocity establisheda general solution involving distributions of three-dimension- andthe plate thickness. The platethickness corresponded to half a wave- al bosses.A prominentadvantage of the theoryover others is that it can lengthbased on theshear wave velocity. The contribution of theringing describebosses of arbitraryorientation. The purposeof the presentpaper wavecomponent to scatteringfrom wet cracks, dry cracks, and ridges are isto comparethe theory with laboratory measurements of sound scattered demonstrated.The impact of thesenew findings on interpreting laborato- by a continuouslyrough pressure-release surface. In the model,we use ry scaledstudies on Arctic acoustics are discussed in viewof thewide use prolatespheroidal Dosses oriented in a rangeof directionsand randomly ofPlexiglas to modelsea ice. Examples are given on scattering from single packedto approximatethe rough surface. The data were taken from Wel- andmultiple cracks and ridges. [Work supported by DREP andONR. ] ton [P. J. Wclton,report A RL-TR-75-30,University of Texasat Austin (1975) ] andinvolve backscattered sound versus grazing angle for several frequencies.The comparisonsof the numericalcomputations with Wel- ton'sdata show a reasonableagreement with respect to notonly the boss 9:20 orientationparameters but alsothe surfacestatistical parameters (rms roughnessand correlation length). An interestingphenomenon isthat the U4. An applicationof a full scatteringtheory to Arctic propagation.J. numberof spheroidalmodes required to fit thedata increases as the fre- S. Hanna, D. M. Rubenstein, L. B. Dozier, and C. Pearson (Science quencyof the incidentplane wave increases. Applications International Corporation, 1710 Goodridge Drive, McLean, VA 22102)
The full scatteringtheory of Burkeand Twersky [J. Acoust.Soc. 40, 883-895 (1966)], which includesboth the specularand scattered 10:05 energyfor a roughsurface consisting of ellipticalbosses on a planesurface, U7. Diffraction of harmonicacoustic radiation by the apex of a ridge. isapplied to theproblem of propagationunder the ice surface. This theory Michael J. Buckingham(Radio and NavigationDepartment, Royal has beenefficiently parametrized [D. Rubenstein(unpublished manu- Aircraft Establishment,Farnborough, Hampshire GUI4, 6TD, script)] for the caseof randomlyspaced keels with a Rayleighdistribu- England) tion of depths.The scatteringkernel for thistreatment has been used in a coupledmode formulationof the ASTRAL propagationmodel. The In general,there are threecomponents in the acousticfield propagat- modifiedASTRAL can computethe propagationassociated with both ingover a ridge:an arrivalvia a directpath, which does not interact with the specularfield and the total fieldwhich includes the nonspecularcom- the ridge,a componentdiffracted by the apexof the ridge,and a compo- ponent.The redistributionof energyamong the modesoccurs continuous- nentreflected from the boundaryof the ridge.All threecontributions are ly with range,so the modelcan display the modeexcitation as a function includedin a new,exact solution of the Helmholtzequation for the three- of range.The modelis comparedto somemeasured propagation data, dimensional,harmonic field from a point sourceover a ridgeformed by paticularlythat of Diachok [J. Acoust.Soc. Am. 59, 1110-1120 (1976) ], two intersecting,pressore-release planes. The solutionis obtainedby the who appliedlow- and high-frequencyasymptotic forms of the Burke- applicationof threeintegral transforms to the Helmholtzequation, fol- Twerskytheory for just the specularfield. [Work supportedby the ASW lowedby the correspondinginverse transforms. It is valid for any ridge EnvironmentalAcoustic Support Program. ] angle0o, including the special case of a knifeedge ( 0o = 360ø) and a plane ( 8o = 180ø).In thelatter case, which is the Lloyd'smirror problem, there is no diffraction;only the directand reflectedcomponents contribute to the field,which is highly variablebecause of strongspatial interference. 9:35 Diffractionby the knifeedge is exemplifiedby the leakageof energyover the barrier,and by the presenceof spatialinterference in regionswhere US. High-frequencyreflection and scatteringby multicomponentrough there is no reflectedcomponent. Similar phenomenaappear with other surfacedistributions. R. J. Lucas'• and V. Twersky(Mathematics ridgegeometries, where it is usuallypossible to assesswhether interfer- Department,University of Illinois, Chicago,IL 60680) enceeffects are due predominantlyto diffractionor reflection. Earlier formsfor coherentreflection and incoherentscattering by mixturesof differentbosses on rigid or freebase planes [V. Twersky,J. 10:20 Acoust.Soc. Am. 29, 209-225 (1957) ] are appliedto recentresults for triaxial ellipsoids[R. J. Lucasand V. Twersky,J. Acoust.Soc. Am. US. Acoustic fields scattered simulated three-dimensional wind-driven $uppl. I 81, :520(1987)] to investigatecontinuous distributions. Key sea surfaces. Jerald W. Caruthers (NAVOCEANO, Oeophysics variablesd (e.g.,keel depth) are governed by two-parameterprobability Department, NSTL, MS 35925), Richard S. Keiffer (NORDA, densityfunctionsf(m, u:d) with ra as the meanvalue of d, and u as the NumericalModeling Div., NSTL, MS 35925), and JorgeC. Novarini normalizedvariance whose value (zero to unity) determinesthe curveof (Syntek Engineering& Computer Systems,Inc., 6110 Executive f For u small,fis Gaussian(and reducesto a deltafunction as u ap- Boulevard, Suite 320, Rockville, MD 20852)
S47 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S47 Statistical characteristics of acoustic fields scattered in the forward rainedfrom volume backscatter. Individual targets that are normally ran- direction from simulated three-dimensional wind-driven sea surfaces are dom and sparserelative to the acousticwavelength, yield echoeswith studiedfor frequenciesfrom 0.1-10 kHz, highgrazing angles, and ranges randomphases between independent realizations. This difficultyhas been in the Fresnel region. The simulated surfaceshave the Pierson- overcomeby complexsuperposition of echoesfrom successive transmis- Moskowitzdirectional spectrum for a 5-m/swind speed. Complex pres- sions,such that the effectivedensity of acoustictargets progressively in- suresare calculated using Helmholtz-Kirchhoff theory with a minimum creasesas the summationproceeds. The theoreticalbasis of this typeof numberof approximations.Statistics of theacoustic fields are calculated coherentprocessing has been developed and examinedin the limiting asensemble averages of the complexpressures and their moments.Com- case,in whichit approximatesa scatterercontinuum. The theoryprovides parisousare presented for differencesin theensemble of surfaces,spectral fundamentalinsights to the behaviorof both the amplitudeand phaseof forms,surface illumination patterns, and implementationof the scatter- compositeechoes from a volumemirror. Experimentalwork in Saanich ing theory.The techniqueallows for the accuraterepresentation of first- Inlet, B.C. servesto confirmthe theoretical concepts through arms differ- and higher-ordermoments of the acousticfield scatteredfrom a three- encemeasurement of theacoustic phase which removes fluctuations from dimensionalsurface. Such quantities are usuallydifficult to obtain by theevolving sound-speed profile. Based on theseresults, the potential for othertechniques. [This work has been supported by NAVOCEANO and newtypes of oceanographicmeasurements using these techniques appears NORDA. 1 both realisticand encouraging.
10'.35 11:20
U9. The effect of finite wind fetch on the acoustic field scattered from U12. Investigation of 12-kHz-deep scattering layers observedwith the wind-drivensea surfaces.Jorge C. Novarini ($yntek Engineering& multibeam echo-sounder Sea Beam. Christian de Moustier (Marine ComputerSystems, Inc., 6110Executive Boulevard, Suite 320, Rockville, PhysicalLaboratory, Scripps Institution of Oceanography,La Jolla,CA MD 20852), Jetaid W. Caruthers (NAVOCEANO, Geophysics 92093) and Dimitri Alexandrou(Department of ElectricalEngineering, Department,NSTL, MS 39529), and Richard S. Keiffer (NORDA, Duke University,Durham, NC 27706) Numerical Modeling Division, Code 221, NSTL, MS 39529) During a surveyover Horizon Guyot in March 1986,acoustic mea- The effectof a finitefetch on the acousticfield scattered by the ocean surementsin the deepscattering layers were carried out with the 12-kHz wind-drivensurface for variouswind speeds is analyzed through numeri- multibeam echo-sounder Sea Beam installed on the R.V. AXLANTm lI. cal simulation.The JONSWAP spectrumis adoptedto describethe sur- Acousticreturns received by eachof the SeaBeam's 16 preformedbeams facestatistics of growingseas and the Pierson-Moskowitzspectrum for weresampled in quadratureat I kHz from the surfacedown to about fully developedseas. Ensembles of thrce-dimensionalsurfaces having the 1400-mdepth, and recorded digitally with a specialpurpose data acquisi- desireddirectional spectra are numericallygenerated. For eachsurface tion systemoriginally developed at the Marine PhysicalLaboratory for realization,the complex pressure of theforward scattered field is calculat- seaflooracoustic backscatter measurements. These complex volume re- ed within the frameworkof the Helmhoitz-Kirchhoff theory for kr> I. verberationdata are presentedas horizontaland verticalcross sections Statistical moments of the scattered field are then calculated as ensemble showingthe 3-D structureof the scatteringlayers observed. Statistical averages.The fieldsscattered in the windwardand cross-winddirections analysesof the datainvolve homogeneity and stationaritytests on a beam are evalutedfor frequenciesup to 1.5 kHz and for anglesaway from to beamlevel as well ason a singlebeam along the ship's track to establish grazingincidence. Results indicate that the scatteredfield is sensitiveto the degreeof patchinessand the corresponding reverberation levels asso- the differencebetween the JONSWAP and the Pierson-Moskowitzspec- ciatedwith the layersobserved. lWork supportedby ONR.] tra. In particular,the total intensityand the coherenceof the scattered field are significantlyaffected by the wind fetch. [Work supportedby NAVOCEANO and NORDA. ]
11:35
UI3. Volume scatteringmeasurements using parametricsonar. James 10:50 M. Syckand CharlesL. Brown (Code 3112 New LondonLaboratory, Naval UnderwaterSystems Center, New Ixmdon,CT 06320) UI0. Resonatorsystems for in situ bubblemeasurements near, and at, the sea surface.Nigel D. Breitz=1 and HermanMedwin (Physics A parametricsonar has beenused to measurevolume scattering Department,Naval PostgraduateSchool, Monterey, CA 93943) strengthat a numberof oceanlocations. The narrowbeamwidth and absenceof sidelobesof the parametricsonar make it an ideal tool for A parallelplate acoustical resonator has beendeveloped to measure measuringvolume scattering strength as a functionof depth.The resolu- bubble densities and their fluctuations near the sea surface and under tion of this device is such that in some circumstances individual scatterers whitecaps.Calibration has been achieved with microbubbleswhose size can be identified.An exampleof sucha situationis presented.The para- (radiusapproximately 50/.t) wasevaluated by speedof risein an aquar- metric sonarsystem is described.Additional setsof examplesfrom near- ium. Using the "sea" generatedin a 20-m-longwater-wind tunnel,the shoreand off-shore areas are presented for both day and night conditions. resonatorperformance has been shown to beindependent of orientation, depth, and seasurface roughness. A seconddevice, using an upward- lookingresonator, is alsobeing developed to investigatelarger bubbles at the surface.In situacoustical devices are compared with remotesensing techniques,and early results of oceantrials are revealed.[ Work support- 11:S0 edby ONR. ] '• OceanAcoustics Associates, 4021 Sunridge Road, Pebble Beach, CA 93953. UI4. A fast and accurate bistatic bottom reverberation model. D. R. Hailer (Defence Research EstablishmentPacific, FMO, Victoria, British Columbia VOS lB0, Canada)
A simplebistatic reverberation model has been developed to model 11:05 low-frequency,wideband, bottom, or surfacereverberation. The model allows for arbitrary separationof sourceand receiverin both rangeand u11. Acousticphuse measurements from •olume backscatter.R. Del Huston ( Arctic SciencesLtd., 1986Mills Road, Sidney,British Columbia depth,and accommodates both source and receiverbeam patterns. Total VSL 3S1, Canada) reverberationis computed by theincoherent summation of the increment- al contributionsfrom each three-dimensionalray path eormectingthe Stableand usefulmeasurements of theorientation of theacoustic sig- source,scattering element, and receiver.Simplified ray pathcalculations nal vectoror acousticphase as a functionof rangeand time can be oh- aremade with eithera constantor a linearsound speed profile, assuming
S48 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S48 sphericalspreading along each ray, andspecular reflection from the flat, Am. Suppl.182, S124(1987) ] hasbeen successfully used to modelrever- horizontalboundary interfaces. Acoustic characteristics of the surface berationfrom explosivesound sources in the TuftsAbyssal Plain of the and bottomare assumedto be completelyrepresented by the reflection NortheastPacific Ocean. Predictions from the simple reverberation mod- lossand scattering strength functions. A realisticrepresentation of gener- el werefound not onlyto be very similarto thoseof the bistaticgeneric alizedbistatic scattering [D. D. Ellis and D. R. Haller, J. Acoust.Soc. sonarmodel but alsoto requiremuch less computation time.
WEDNESDAY MORNING, 18 MAY 1988 WEST BALLROOM (A&B), 9:00 TO 11:35 A.M.
Session V. Architectural Acoustics II and Musical Acoustics II: Concert Hall Acoustics Based on Ando's Work
Richard Talaske, Chairman Talaske-JoinerGroup, Incorporated, 137 NorthOak Park ,•venue,Oak Park, Illinois60301
Chairman's Introduction--9:00
Invited Papers
9:05
V1. Furtherinvestigation on concerthall acoustics.Yoichi Ando (Facultyof Engineering,Kobe University, Kobe 657, Japan) Sincethe book [Y. Ando, ConcertHall Acoustics(Springer, New York, 1985)] waspublished, further workhas been carried out. The resultsdemonstrate that therunning autocorrelation function of sourcesignals maybetter describe preferred conditions of soundfields than does the long-time autocorrelation function. The minimumvalue of the effectiveduration of the runningautocorrelation function of musicsignals for the intervalused for the subjectivepreference judgment corresponds well to the preferredinitial time delaygap betweenthe direct soundand the first reflectionwhen the amplitudeA • = 0 dB. This wasdiscovered when Japanesemusic containing a so-called"MA," a stationarysound passage without any rapidmovement, was usedas a sourcesignal. Also, it wasfound that a flat frequencycharacteristic of the subsequentreverberation timeis the optimal condition, and that it maybe calculated by the minimum effective duration of therunning autocorrelationfunction of the sourcesignal. Several other results will be presented.
9:35
V•2.Early reactionsto YoiehiAndo's book, Concert Hall Acoustics.Leo L. Beranek(BBN Laboratories,10 Moulton Street,Cambridge, MA 02238) Ando'sbook joins one of a handfulwritten since 1900 on the scienceof concerthall acoustics.It contains sevenchapters packed with physical acoustics, the behavior of the physical hearing system, the response ofthe nervoussystem, psychoacoustic judgments, the composition of a soundfield in a hall,prediction of subjective preferencesin halls based on the sound field, and design studies for concert halls. This paper attempts to relate Ando'steachings to the practical experience the author has had with the acoustics of a numberof concert halls. Areas in which the book fits theseexamples well and in whichfurther researchseems indicated will be presented.
10:05
V3. Solvedand unsolvedproblems in concerthall acoustics.Manfred R. Schroeder(Drittes Physikalisches Institut, Universityof Goettingen,D-3400 Goettingen,Federal Republic of Germany) Duringthe last2 decades,several difficult problems in concerthall acousticshave been solved or elucidat- ed.Among these, the recognition of theimportance of earlylateral sound ranks high. This pieceof progresswas made possiblcby the combinedefforts of expertsfrom widely varying disciplines(physicists, psychologists, mathematicians, and--last but not least--the observant consultant). But much remains to be done to strengthenthe scienceof room acousticsfurther. With digitalelectronics reaching new pinnacles of sophisticaq tion,active soundfieM modification should be pursuedmuch more actively--for multipurpose halls and other spaces.Diffusion of radiatedsound can be controlledby the samenumber-theoretic principles that are finding
S49 J. Acoust.Soc. Am. Suppl;1, Vol. 83, Spring1988 115th Meeting:Acoustical Society of America $49 increasinguse in passivedesigns [ M. R. Schroeder,Number Theory in Science and Communication (Springer, NewYork, 1986),2rid ed. ]. Butthe optimum degree of diffusionrequires further subjective studies. Generally usablesoftware for digital modeling of acoustic spaces should be developed toexploit more fully the capabilities of moderncomputers (both serial and parallel). Room acoustical measurement methods that use speech and musicas test signals, to permitmeasurements during actual performances, should be perfected and used as a matterof course.The future of roomacoustics isbright if wegrasp the available opportunities.
ContributedPapers 10:35 10:50
V4. Soundfield modelingand simulationof a concerthall. T. Hidaka, V5. Effect of transient signal length on interaural cross-correlation K. Kageyama, and S. Masuda (Takenaka Technical Research functions. Hirofumi Yanagawa (Department of OMDD, Pioneer Laboratory5-14, 2-chome, Minamisuna, Koto-ku, Tokyo 136, Japan) ElectronicCorporation, 2610 4-chome,Hanazono, Tokorozawa City, Saitama359, Japan) A soundfield simulator has been developed in orderto subjectively evaluatehall acousticsin the laboratory.This processconsists of three parts:( 1) a computerprogram for derivingthe impulseresponse of the The internalcross correlation is closelyrelated to thesubjective im- roomfrom the architectural drawings using the ray tracing technique; the pressionsofsound fields. This paper attempts to account, in a comprehen- programis writtenmainly in FORTRANand the calculationis implemen- sive way, for the combinedeffects that initial reflectedsound and rever- ted usinga maximumof ] 500 planesurfaces and I 000 000 raysat most; berationsound from musicor other transientsignals have on such (2) a newsource of orchestralmusic that is digitally recorded in a tempo- impressions.To this end, cross-correlationfunctions of the riseand fall of rarily constructedanechoic room for this purpose,and (3) hardwareto thesound field from transient signals were derived from the windowing synthesizethe calculatedsound field, which convolutes impulse response impulseresponses at two pointsin a hall. Theseresults were combined with inputsignal by digitalsignal processing and outputs from 24 speak- withthe image-sources distribution patterns derived by the closely locat- ersset up in an anechoicroom in realtime. In orderto applythis simula- ed four-pointmicrophone method. Good agreement was found between tor, subjectiveexperience by meansof auditionfor comparisonof sixcon- changeswith time in experimentallyderived maximal cross-correlation certhalls is madeand its abilityand accuracy are investigated. functionvalues and the image-sourcesdistribution of the soundfield.
11:05-11:35
Panel Discussion
PANELISTS: Yoichi Ando Leo L. Beranek Manfred R. Schroeder
WEDNESDAY MORNING, 18 MAY 1988 GRAND BALLROOM B, 9:30 TO 11:15 A.M.
SessionW. PsychologicalAcoustics III: Discriminationand Perception of ComplexStimuli
Robert S. Schlauch, Chairman PsychologyDepartment, University of California,Berkeley, California 94720
ContributedPapers
9:30 controlledby meansof a cuemelody preceding the embeddedtarget. In W1. Directingattention in pitchand time: Effects on the perceptionof halfthe sessions,the cue indicated the pitch level of thetarget, and listen- hiddenmelodies. W. Jay Dowling(Program in HumanDevelopment ersjudged the time at whichit occurred.In the othersessions, the cue and CommunicationSciences, University of Texas/Dallas,Richardson, indicatedtime, and listenersjudged pitch. The cuewas 83% valid, and TX 75083-0688) expectedtargets matched the cue while unexpected targets did not. Rate of presentationwas adjusted within each session to produceapproximately Listenersattended to a targettone in themiddle of a familiarmelody 80% accuracy.Temporal expectantics had a muchgreater effect on reac- that wasinterleaved with distractornotes. The targetoccurred in either tion times(RTs) than did pitchexpectantics. RTs for pitchjudgments expectedor unexpectedlocations in pitch or in time. Expectancywas wereabout 130 ms longerwhen targets occurred at unexpected(versus
$50 J.Acoust. Soc. Am. Suppl. 1, Vol. 83, Spdng 1988 115thMeeting: Acoustical Society of America $50 expected)times, while RTs for timejudgments were only about25 ms animal'ssensitivity to ripplednoise was a functionof thecenter frequency longerfor targetsat unexpected(versus expected) pitches. Rhythmic of thenoise. The dolphin'sperformance was relatively constant as a func- organizationprovides a basisfor the listenerto directattention to critical tion of centerfrequency. The overallresults suggest that dolphinsmay be eventsin a rapid auditorystream. ableto perceivetime-separation pitch.
9:45 10:30
W2. The subjective size of musical inter•als. Lync Platoondon W5. Differential frequencysensitivity of humansand monkeysfor tone (Department of Psychology,University of California, Berkeley,CA versus vowel stimuli. Joan M. Sinnott (Department of Psychology, 94720) IndianaUniversity, Bloomington, IN 47405)
The musicalscale and the psychephysical scales of pitchappear to be Humansand monkeys (Macacafuscata) wcrc compared in theirabili- in contradiction with one another. The debate centers on whether or not tiesto discriminatefrequency changes (a) in a 2-kHz puretone and (b) pitch is a log functionof frequency.Previous results on the listeners' alonga steady-state/I/to/i/vowel continuum(F1 = 390- 270 Hz; judgmentof ratiosand differencesin pitch betweenpairs of tonesagree F2 = 1990-2290Hz; F3: 2550-3010 Hz). A repeatingstandard AX with themusical scale JR. Elmasianand M. H. Birnbaum,PercePt. Psy- procedurewas used and stimuli were presented monaurally through the chephys.36, 531-537 (1984) ]. In the presentexperiment, the subjective rightcar. For thetone stimulus, human and monkey DLs weremarkedly sizeof musicaland nonmusicalintervals was measuredby magnitude divergent;at lowSLs, the human DL (6 Hz) wasabout 1/5 thatof mon- estimationand categoryrating, usingsubjects without muscial training. keys (30 Hz) while, at high SLs, the humanDL (4 Hz) decreasedto Pure tonesas well as complextones (harmonic series)were used.Fre- about 1/20 that of monkeys(80 Hz), as monkeysbecame less sensitive. quenciesranged from 110-2145 Hz. For each pair of tones presented For the vowel stimulus,human and monkeyDLs (measuredrelative to melodically,subjects were instructedto rate the pitch of the secondtone changesin F2) weremore similar: The humanDL ( 15Hz) wasabout 1/3 relativeto the pitch of the first one. Resultsshow that subjectsuse the that of monkeys(45 Hz), andSL functionswere parallel and fiat for both samecomparison operation for both ratio and differencejudgments of species.Results indicate that monkeysprovide a muchbetter (although pitch.The effectsof frequencyrange, spectral composition, size of inter- not perfect)model of humansensitivity for a complexvowel stimulus val, andmusicality of the ratioson the subjectivesize of musicalintervals than for a simpletone stimulus. [Work supportedby NIH.] are discussed.[Research supported by N IH.]
10:45 10:00 W6. Speechperception by budgerigars(Melopsittacus undulatus): Spoken W3. Perceptionof timing in rhythmic patterns.Caroline B. Monaban vowels. Susan D. Brown, Robert J. Doeling, and Kazuo Okanoya and Ira J. Hirsh (Central Institute for the Deaf, St. Louis, Me 63110) (Departmentof Psychology,University of Maryland,College Park, MD 20742) In twostudies employing an adaptive,cuecl-2^FC procedure, listen- ersdiscriminated between six-tone rhythmic patterns that differedonly in Budgerigars(parakeets) were trained on a same/differenttask using the delayof the temporalposition of one of the tones.On eachtrial, operantconditioning. Response latencies were used to constructsimilar- feedbackwas given and the subject'sperformance determined the amount ity matrices,and multidimensionalscaling procedures were then used to of delayon the next trial. The six tonesof the patternsmarked off five producespatial maps of thesespeech stimuli. Budgerigars were tested on intervals.In the firststudy, eight patterns comprised three "short" and stimulussets consisting of naturalexemplars of differentvowels produced two "long" intervals:thus 12121,21121, etc., where the "long" (2) was by a numberof male and femaletalkers. In spiteof variationamong twice the lengthof a "short" (1). In the secondstudy, eight patterns talkers,budgerigars showed evidence of perceptualcategories for differ- comprisedtwo "shorts"and three"longs": thus 21212, 122! 2, etc. Each ent vowels as well as for male and female talkers. The two-dimensional pattern was tested45 times Ifive positionsof the delayedtoneXthree perceptualspace for budgerigarsdiscriminating among vowelsbears a temposX three replications)for eachof threesubjects. Consistent with strikingresemblance to the familiarF I-F2 vowelspace described by Pe- previouswork on simpleinterval discrimination,absolute discrimination tersonand Barney [J. Acoust.Sec. Am. 24, 175-184 (1952)]. This sug- (At in ms) waspoorer, the slowerthe tempo.Relative discrimination geststhat budgerigarsuse first and secondformant information in discri- (A[/t "short," where "short" was 50, 100, or 200 ms) was better, the minatingamong spoken, sustained vowels. [Work supportedby NIH.] slowerthe tempo.Beyond these global results, large interactions of pat- tern with positionof the delayedtone and temposuggest that different modelsof performanceare neededto explainbehavior at the different tempos.[Work supportedby AFOSR.] 11:00
W7. Speech perceptionby budgerigars(Melopsittacus undulatus): SyntheticVeT stimuli.Robert J. Doeling,Kazuo Okanoya,and Susan 10:15 D. Brown(Department of Psychology,University of Maryland,College Park, MD 20742) W4. The perceptionof time-separationpitch by dolphins.Whitlow W. L. Au (Naval Ocean SystemsCenter, P.O. Box 997, Kailua, HI 96734) Budgerigars(parakeets) were trained using operant conditioning pro- and Jeffrey Paw]oski(SEACO Inc., Kailua, HI 96734) ceduresto detectchanges in a repeatingbackground of syntheticspeech sounds.Response latchtics for detectionwere used to constructsimilarity A dolphinwas required to discriminatebetween rippled and nonrip- matrices,and multidimensional scaling procedures were then used to pro- pled noiseprojected by an underwatertransducer. Random noise was ducespatial maps of thesespeech stimuli reflecting perceptual organiza- summedwith its delayed replica to producenoise having ripples separated tion. Birds were testedon the samealveolar, velar, and bilabialsynthetic by 1/T Hz in the frequencydomain, where Tis the delaytime. The dol- VeT stimulipreviously used to testhumans and chinchillas [ P. K. Kuhl phin detectedthe cos- rippledstimulus at a correctresponse level of at and J. D. Miller, J. Acoust. Sec. Am. 63, 905-917 (1978) ]. Resultsindi- least75% for delaysbetween 15 and 500#s, andthe cos+ rippledstimu- catethat budgerigarspartition each of thesecontinua into two perceptual lus for delaysof 13 to 190/•s. The dolphin'ssensitivity to ripplednoise was categorieswith an abrupt perceptual change occurring at roughly the measuredby attenuatingthe delayedreplica for differentdelays. The dol- same VeT location as observed in humans and chinchillas. It is concluded phinwas most sensitive for a delayof 100/zs.Its sensitivityat 100•ts was5 that the perceptualmechanisms underlying the categorigalperception of dB better for the cos + than the cos- stimuli. The broadband cos VeT stimuli are sufficientlygeneral so as to be commonto the mamma- • noise was also filtered in different 1/3 octave bands to determined if the lian and arian auditorysystems. [Work supportedby NIH.]
S$1 J. Acoust.Sec. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America $$1 WEDNESDAY MORNING, 18 MAY 1988 METROPOLITAN BALLROOM, 10:00 A.M. TO 12:00 NOON
SessionX. SpeechCommunication IV: Neural Networksin SpeechRecognition II {PosterSession}
Les Atlas, Chairman Departmentof ElectricalEngineering, FT-10, Universityof Washington,Seattle, Washington 98195
ContributedPapers All posterswill bedisplayed from 10:00a.m. to 12:00noon. To allowcontributors the opportunityto seeother posters,contributors of odd-numberedpapers will be at their postersfrom 10:(30to ! 1.00a.m. and contributors of even-numberedpapers will be at their postersfrom I 1.00a.m. to 12:00noon.
Xl. A studyon the speaker-independentfeature extractinnof Japanese Four two-layerassociative networks (one layer of hiddennodes) were vowelsby neuralnetworks. Toshio Irino and Hideld Kawahara (NTT trainedto classifyspectra extracted from naturally produced vowels of BasicResearch Laboratories, 3-9-11 Midori-cho Musashino, Tokyo 180, onetalker, using the hack-propagation learning algorithm [ Rumelhartet Japan) ai., Nature:523, 533-536 (1986) ]. The inputsto thenetworks were either LPC spectraor "perceptual"spectra, obtained by convolvingthe Bark- The featureextraction characteristics of three-layerback propagation transformedspectrum with an asymmetric filter. The networks trained on neuralnetworks, when applied to speaker-independentvowel/gender rec- LPC input classifiedtest vowelsmore accuratelythan the networks ognitiontasks, are investigated. Speech samples are monosyllabicvowels trainedon the perceptuallybased spectra. The networks'solutions were extractedfrom a syllablicdata base of IO0Japanese syllables digitized at a examinedto determine(a) whetherit was possibleto characterizethe samplingrate of 16kHz spokenby 100different speakers. Several LPC- featuresunderlying the solutions and (b) whetherthe input spectral rep- basedparameters and physiology-basedparameters are testedfor input resentationaffected the nature of the solutions.Principal components repr-c•entations.The resultsreveal that three hidden units are necessary to analysesof theactivation patterns of thehidden units showed that three discriminatefive Japanese vowels. Close investigation of hiddenunit func- dimensions accounted for over 80% of the variance for all networks. For tionsreveals that distributedrepresentations of the targetsare developed networkstrained on "perceptual"spectra, the first three dimensions were as hiddenunit activationpatterns. The frequencydomain interpretation highlycorrelated with F2/F 1ratio, F 3/F 2ratio, and F 3,respectively. For of theinput weights of hiddenunits, using autocorrelation inputs, shows LPCinput, the first t h ree dimensions correlated with F 1,F 3/F 2 ratio,an d that the network extractsconventional knowledge on vowel formant F2/F 1 ratio.Thus these networks classified the vowelsusing features re- structure.The relationsbetween hidden unit activationpatterns and de- lated to the formant space. scriptivefeatures, as well as their implicationsto speechperception re- search,are also discussed.
X4. Completegradient optimization of a recurrentnetwork applied to X2. Reeoguitionof vowels using artificial neural networim. Hong fo/, /d/, /g/ discrimination.Raymond L. Watrous (SiemensRTL, 105 C. Leungand Victor W. Zue (Room36-547, Department of Electrical College Road East, Princeton,NJ 08540 and the Universityof Engineeringand Computer Science,Massachusetts Institute of Pennsylvania,Philadelphia, PA 19104), Bruce Ladendoff (Siemens Technology,Cambridge, MA 02139) RTL, 105 CollegeRoad East, Princeton,NJ 08540), and Gary Kuhn (IDA-CRD, 1 Thanet Road, Princeton, NJ 08540) This paperis concernedwith the applicationof artificialneural net- worksto phoneticrecognition. This work is motivated by theobservation A completegradient optimization technique for connectionistnet- that improvedknowledge of featureextraction is oftenovershadowed by workswith recurrentlinks is presented. A truncatedgradient technique is relativeignorance on how to combinethem into a robustdecision. The available in the literature. A network with recurrent links for discrimina- goalis to investigatehow the self-organizingframework of artificialneu- tion of fo/,/d/, and/g/in the contextof following/i/,/a/, or/u/is ral networkscan be exploited to enabledifferent acoustic cues to interact. designed.The performanceof thisnetwork as optimized with either of the The investigationis couchedin experimentsthat recognizethe 16 vowels two techniquesis then compared. It will beshown that thecomplete gradi- in AmericanEnglish, using some 10 000 tokensin all phoneticcontexts. ent isexact and simpler, and it will bedemonstrated that it leadsto superi- The tokenswere extracted from 1000sentences spoken by 140males and or performance. 60 females.It wasfound that, by replacingthe mcan-squared error metric with a weightedone to train a multilayerperception, better recognition accuracyand rank orderstatistics were consistently obtained. Using the two-layerperceptron in a context-independentmanner, a top-choiceaccu- racyof 54% wasachieved, which compares favorably with resultsreport- X$. Speechestegurization using recurrent networks. Sven Anderson, ed in the literature. Context-dependentexperiments reveal that heterogen- John Merrill, and Robert Port (Department of Linguistics,Indiana eoussources of informationcan be integratedto improverecognition University,Bloomington, IN 47405) performance.The top-choiceaccuracy of 67% is comparableto the aver- Severalconnectionist networks were trained to classi• the English ageagreement among listeners. Finally, it wasfound that the rateof im- syllablesha, ds. ga, pa, ta, ka collectedfrom two male and two female provementon recognition accuracy may be used as a terminatingcriterion speakers.Using a speechpreprocessor, perceptually based spectral pat- fortraining, and that reasonable performance can be achieved using as few terns were computed[I-L Hermansky,Proc. ICASSP 87, 1159-1162 as 800 trainingtokens. [Work supportedby DARPA.] (1987) ] every5 ms.A sequentialnetwork having a limitedclass of recur- rent connections[M. Jordan,ICS Tech. Rep. Universityof Californiaat SanDiego (1986) ) wasemployed to categorizethe data. Training by back propagationor second-orderback propagation, a linear increase in the X3. A e0de in the nodes, Lynn A. Streeter, Candace A. Kamm certaintyof classificationover the courseof the syllablewas required. (Bellcore,435 SouthStreet, Morristown), NJ 07960), and Yana Kane- Performanceof thesequential networks was evaluated on both "known" Esrig(Cornell University, Ithaca, NY 14850) and "unknown"speakers. When testedon novel tokensof a known
S52 J. Aoou-•t.Soc. Am. Suppl. 1, VoL83, Spring1988 115thMeeting: Acoustical Society of America S52 speaker,the sequential network did very well as oppsed to verypoorly on Thegoal here is to isolatethese factors. In thepresent contribution, differ- tokensfrom an unknownspeaker. Sequential networks trained with back entfront ends in theneural net-based ASR aresystematically evaluated. propagationare capable of integratingcues distributed over time and us- Severalstandard ASR frontends are compared with the synchrony audi- ing them to categorizedata. However,their learnedbehavior may not tory modeland the perceptuallybased linear predictive auditory model generalizeto datafrom other speakers. [Work supportedby NSF.] frontends in boththe speaker-dependent andthe speaker-independent ASR.The speed of learning and the ASR accuracy of thecompared recog- nizersare reported and discussed.
X6. Effectsof networktopology on speech categorization. John Merrill, SvenAnderson, and Robert Port (Departmentof Linguistics,Indiana University,Bloomington, IN 47405) X9. Isolated word recognitionbased on the adaptiveneural network.J. Recognizingspeech requires the identification and anlysis of tempor- S. Wang and B. Widrow (ISL, Stanford University, Stanford, CA 94305 ) ally distributedcues. A systemmust either examine a sufficientlylong windowat a singleglance or elseinternally accumulate stimulus informa- For the last severalyears, despite many new approaches(such as tion.Sequential networks follow the second path by storinginformation applicationof the Markov'model and VQ) in speechrecognition, the internally in statenodes. Feed-forward networks do not maintain their classicalpattern matching using dynamic programming still prevailsbe- historyinternally and thus require that thespeech signal be presented in causeit yieldsgood recognition results. However, this DP-basedalgo- fixedwindows. The performancesof sequentialand feed-forwardnet- rithm requiresa largeamount of computation.It is difficultto meetthe worksat recognitionof auditorilypreprocessed stop-vowel syllables are real-timerequirement for largevocabularies. An alternativepattern rec- compared.Several feed-forward networks were trained by presentinga ognitionmethod called the adaptiveneuron in adaptivecontrol system wholesyllable to thenetwork as a singletoken and requiring categoriza- usedby B. Widrow in the early 1960sis usedhere to buildan isolatedword tion.The sequential networks were more robust within and across speak- recognitionsystem. A simpleand efficient learning algorithm is presented ersthan the feed-forwardnetworks. Unfortunately, the useof the back- for adaptivelyadjusting weight vectors to fit a certainword pattern. In propagationalgorithm to train a sequentialnetwork requires presentation order to capturethe variationsin speech,an increasingcodebook was of a desiredoutput at eoerytime slice. This forcesarbitrary choices for designedduring the learningphase to allow increasinglymore patterns; specifyingtarget outputs. The resultssuggest that techniquesbased on therefore,multiple referencepatterns were adopted,giving the advan- back-propagationwill proveinadequate to train networksto perform tagesof a statisticalMarkov model.The performanceof this recognition speechcategorization across variations in speaker.This difficulty could be systemhas been preliminarily tested on twocommon vocabularies includ- obviatedby employmentof a learningparadigm that doesnot require ingten digits and the English alphabet. The recognitionaccuracy is 100% immediatefeedback, such as a stochasticlearning algorithm. [ Worksup- for digitsafter five trainings per word and 93% for the alphabetafter ten portedby NSF. ] trainingsper word for speakertrained. Further experimentsfor a com- pleteChinese vocabulary are underprogress.
X7. Usingan adaptivenetwork to recognizedemisyllables in continuous speech.Candace A. Kamm,Thomas K. Landauer,and Sharad Singhal X10. An artificial neural network-basedcodebook search technique. (Bellcore,435 SouthStreet, Morristown, NJ 07960) Charles Pope, Les Atlas, and Charles Nelson (Department of A nonlinear,multilayer associative network was trained on a speech ElectricalEngineering, FT-10, Universityof Washington,Seattle, WA recognitiontask using continuous speech. Naturally spoken 14-syllable 98195) "sentences"from one talker were preprocessedto producea 15-band An unsupervisedlearning artificial neural network (ANN) IT. Ko- spectralrepresentation incorporating several transformations introduced honen,Helsinki University of TechnologyReport TKK-F-A601 (1986) ] bythe peripheral auditory system on acoustic signals. Input nodes to the wasmodified for a vector-quantizedPCM (VPCM) codebooksearch networkrepresented a 150-ms window through which the spectral repre- problem.The ANN wasstructured such that each element of thenetwork sentationpassed in 2-mssteps. A singlelayer of 20hidden nodes was used. hadassociated with it a singlecodebook vector. The amountof processing Outputnodes represented seven initial demisyllables whose target values requiredat eachelement of the networkwas usedto derivean upper werespecified based on a humanlistener's identification of the sounds boundon the numberof networkiterations. Simulations were performed heardduring the input segment.The networkwas trainedto criterion to determinethe effectthat variousnetwork parameters had on the speed usinga variantof theback-propagation learning algorithm [Rumelhart et of networkconvergence. The parametervalues offering the greatest per- al., Nature323, 533-536 (1986); Landaueret al., Proc.Cog. Sci. Soc., formante were appliedto the network.The speedand computational 531-536 (1987) ]. A minimum-error-ratefigure of merit (derivedfrom complexityof the ANN solutionto thisproblem were then compared to signaldetection theory) wasused to evaluatethe effectof the sizeof the the samecriteria for a standardlinear (full-codebook)search technique. trainingcorpus on the network'sperformance. Minimum error rate on a Analysisand test results indicate that theANN approachcan provide the testcorpus of demisyllablesspoken in differentcontexts decreased from speedof a treesearch coupled with the minimummemory characteristics 5.2% to lessthan 2% as the numberof sentencesin the trainingcorpus associatedwith a linearsearch, at the expenseof requiringa multiproces- was increased from one to seven. sorconfiguration. [Work supportedby NSF.]
XS. Comparisonof standard ASR front ends and auditory models in Xll. Enhanced speaker identification through the use of cepstral neural net-based automatic speech recognition. Mark Terry and coefficientsand redundant time frameelimination. Timothy R. Thomas, Hynek Hermansky (U.S. West AdvancedTechnologies, 6200 South GeorgeJ. Papeun(Los AlamosNational Laboratory, Los Alamos, NM QuebecRoad, Engelwood,CO 80111) 87545), Start Willie, and Kalyan Ganesan (U.S. WEST Advanced Technologies,Englewood, CO 80111) Recentwork [Renalsand Terry, EuropeanConference on Speech Technology,Edinburgh, United Kingdom(1987) ] reportedon a small A text-independentspeaker identification system was developedon multispeakerisolated digit automaticspeech recognition (ASR) experi- two maleand two femalespeakers with similaraccents. Each read a pho- mentusing a backwardpropagation neural net systemwith a synchrony neticallybalanced list often sentences.The systemwas trained repeatedly auditorymodel as its frontend. In spiteof fairly crudetemporal normali- on a rotatedset of ninesentences and tested on the remainingone. Speech zation,the systemwas capable of betterthan 80% ASR accuracy.The duringsuccessive 16-ms windowed time sliceswas described by 14 eep- questionremains if the reportedperformance is due to (a) the paficular stral coet•cients. Unit direction vectors were used to characterize each front end, (b) the particularneural net-based classification, (c) the par- sentencefrom each speaker. A nearest-centroid,nearest-neighbor, or im- ticular time-normalization scheme,or (d) the combination of all factors. provedperceptron neural net trainingprocedure was used to definedeei-
853 d. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S53 sionregions. When the datawere preprocessed so as to removetime slices resultprimarily from the ability ofthe cepstral coefficients tocapture the that weresimilar in all speakers,discriminability was enhanced and error- speaker-dependentinformation in the higherformants and from the ac- lessidentification was obtained.The successof this systemappears to centuationof this informationby the preprocessor.
WEDNESDAY AFTERNOON, 18 MAY 1988 GRAND BALLROOM C, 12:45 TO 2:32 P.M.
SessionY. SpeechCommunication V: Speechand Speaker Recognition
Victor W. Zue, Chairman Departmentof ElectricalEngineering and ComputerScience, Room 36-591, Massachusetts Institute of Technology, 77MassachusettsAvenue, Cambridge, Massachusetts 02139
Chairman's Introduction•12:4.•
ContributedPapers
12:47 tal frequencyinformation is then used to reclassifyuncertain cases as eitherstressed or unstressed.Preliminary evaluation of the system'scur- Y1. Comparativestudy of ASR front-endsin noise.Jcan-Claude Junqua rent performanceon a multispeakerdata base yielded 77% correctloca- (SpeechTechnology Laboratory, 3888 StateStreet, Santa Barbara, CA tion and classificationof syllables.Although improvement is necessary, 93105 and C.R.I.N.-CN.R.S., Vandoeuvre les Nancy, France) and this resultis encouraging. Hisashi Wakita (SpeechTechnology Laboratory, 3888 State Street, Santa Barbara, CA 93105)
In automaticspeech recognition (ASR) of speechcorrupted by noise, the performancetends to deterioraterapidly depending on the choiceof 1:17 analysismethod and distance measure. In orderto evaluatethe recogni- tion performancefor severalanalysis methods and distancemeasures, a Y3. A segmentationalgorithm based on spectralvariance. A. Kumar and seriesof isolatedword recognition experiments was performed. Analysis H. Wakita (SpeechTechnology Laboratory, 3888 State Street, Santa methodsselected are critical-band filtering, perceptually based linear pre- Barbara, CA 93105) diction(PLP), linearprediction (LP), andtime synchronous linear pre- diction(SLP). The weightedEuclidean distance with differentweight- Presentedisa segmentationalgorithm based on spectral variance. The ings [unity, root power sums(RPS), and exponentialfiltering] was speechsignal is firstsegmented into spectrallystable segments by usinga appliedin the cepstrumdomain. Experiments were carried out for clean mediansmoothed spectral variance over a 70-mswindow. The segment speechand for two noiseconditions (white and low-passfiltered white, boundariesare placedat the maxima in the spectralvariance and the addedto the cleanspeech) at differentSNR ratios(25 to 5 dB), usingan minima givetypical framesfor the segments.The spectralvariance peak for the glidesis generallyvery small becauseof their smoothtransition. alphanumericvocabulary (ten speakers).It is shownthat improvements Hence,the glidesare segmented by usingsecond and third formanttrajec- in robustnessof the recognizerin noisecan be achievedby a properselec- tion of analysismethod and cepstralweights used in the front-end.Im- tories.Extraneous segments outside the word boundariesarc eliminated provementsare foundover the RPS distancemeasure (previously shown by adaptivesilence detector. The segmentsarc then assigned broad phon- to be usefulin noiseconditions with LP and PLP analyses)lB. Hanson emeclasses by usinga tree classifieron the LPC cepstralcoefficients for and H. Wakita, ProceedingsIC,4SSP 86 (IEEE, New York, 1986), pp. the typicalframes. The fricativeand nonfricative segments are distin- guishedby the normalizeddifferenced speech signal. The tree classifieris 757-760] by useof the generalexponential lifter. separatelytrained on hand-labeleddatabases. Preliminary experiments show that 90% of the segmentsare detected.Results for the 104-word keyboard vocabularyfor six males and four femalesand continuous speechwill be presented. 1:02
Y2. Feature-based automatic syllable and stress detection. Briony Williams and JonathanDalby (Centre for SpeechTechnology Research, University of Edinburgh, 80 South Bridge, Edinburgh EH1 1:32 1HN, Scotland) Y4. Speechrecognition using a synthesizedcodebook. L. A. Smith, B. The importanceof syllablestructure and stresslevel as determinants L. Scott, R. G. Goodman, L. S. Lin, and J. M. Newell (Scott of segmentaltemporal and spectralvariability makes automatic syllable InstrumentsCorp., 11! 1 Willow SpringsDrive, Denton,TX 76205) detectionand stressestimation a verydesirable goal for continuousspeech recognitionresearch. In this paper, a rule-basedsystem is describedfor Speechsounds generated by a simplewaveform synthesizer were used locatingsyllables in continuousspeech and for makinga two-levelstress to createa vectorquantization codebook for usein speechrecognition. assignment.Location of syllablenuclei and rough estimationof syllable Recognitionwas testedover the TI-20 isolatedword databaseusing a boundariesare performedusing a smoothedmidfrequency "SOhorant" conventionalDTW matchingalgorithm. Input speechwas filtered to lim- energy contour, a frication detector, and Sohorantconsonant detectors. it the bandwidthto 300-3300 Hz, then was passedthrough the Scott First passclassification of detectedsyllables as stressed,unstressed, or InstrumentsCoretechs process, implemented on the SI2010 signalpro- uncertain is basedon the relative energy levels and utterance-position- eessingchip, to creat• the speechrepresentation for matching.Synthe- normalizeddurations of syllablesin a three-syllablewindow. Fundamen- sizedsounds were processed in softwareby an SI2010emulation program.
S54 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S54 S12010emulation and recognitionwere performed on a DEC VAX 11/ motivatedby the beliefthat thisprocedure may help to identifyacoustic 750. The originalcodebook contained 109 vectors.This codebookwas attributesand decisionstrategies that are usefulfor systemimplementa- decimatedthrough the courseof theexperiments, based on the numberof tion.Listening and spectrogram reading tests involving eight listeners and timeseach vector was used in quantizingthe training data for theprevious sixspectrogram readers, respectively, were conducted using a corpusof experiment.Recognition scores are presentedfor progressivelysmaller 1000wordlike strings designed to minimizethe useof lexicalknowledge. codebooksizes, as well asfor the baselinecondition (no vectorquantiza- Resultsshow that listeners'performance was better than readers' (98.4% tion). vs 91.0%). In both experiments,string lengthswere determinedvery accurately(98.1% and 96.2%), presumablydue to the largenumber of glottalstops inserted at letterboundaries to facilitatesegmentation. Most of the errors were due to substitution of one letter for another (68% and 92%), andthey generally fall into two categories.Asymmetric errors can 1:47 oftenbe attributed to subjects' disregard for contextual influence, whereas YS, An efficient, robust speaker-independentalgorithm. Brian Scott, symmetricerrors are largelydue to acousticsimilarities between certain Lisan Lin, Mark Newell, and Lloyd Smith (Scott Instruments letterpairs. Subsequent acoustic study of four of the mostconfnsable Corporation,1111 Willow Springs Drive, Denton, TX 76205) letterpairs has resulted in theidentification ofa numberof distingnishing acousticattributes. Using these attributes, overall recognition perfor- The algorithmsdescribed have yielded speaker-independent scores of mancebetter than that of the readerswas achieved. [Work supportedby 95.1% on the 20-word TI database obtained from the National Bureau of NSF and DARPA under contract N00014-82-K-0727, monitored Standards.Results were obtained by trainingthe systemon half of the throughthe Office of NavalResearch. ] speakersin thedatabase, testing on theother half, and then reversing the order.Training was done with the 10training tokens per speaker per word only.Testing was on the 16 testtokens per speakerper word.The total numberof testtrials was 5120. The recognizeruses conventional methods 2:17 for time normalizationand matching.Time normalizationis linearand scoringis accomplished with a simpledifferencing algorithm weighted by Y7. Humanand machineperformance on speakeridentity verification. variances.Storage requirement is 3072bits per word. Most of thespeaker TimothyC. Feustel(Bell CommunicationsResearch, 435 SouthStreet, normalizationis accomplished by the proprietary signal processing meth- MRE 2E 236, Morristown, NJ 07960), Robert J. Logan (SUNY at oddeveloped by Scott Instruments. Aside from the amplitude normaliza- Binghamton,Binghamton, NY 13901),and GeorgeA. Veiins(Bell tionroutines, no floating point arithmetic is used. All signalprocessing is CommunicationsResearch, 435 SouthStreet, MILE 2E 236, Morristown, temporallybased. The frontend process can be adapted for usewith dy- NJ 07960) namictime warping algorithms or featurebased algorithms. The system is, therefore,extensible to connectedspeech. Two experimentswere conducted to identifyacoustic features for speakeridentity verification (SIV) thatare used by humansand not by eepstral-basedalgorithms. Although these algorithms generally out-per- formhuman listeners for randomlyselected comparisons between single- word utterances,this approach was to analyzehuman performance on comparisonsthat could not be effectively discriminated by machine. Ex- perimentI showed that humans could perform at highlevels of accuracy 2.'02 on thesecomparisons suggesting that either information exists that is not capturedby the algorithms, or thatthe information iscoded by the algor- Y6. Reco•ition of continuouslyspoken letters by listenersand ithmsbut is not used effectively. The second experiment consisted of three spectrogramreaders, Nancy A. Dalyand Victor W. Zue(R0om 36-575, stimulusconditions for SIV; digitizedspeech signals, noise-excited re- Department of Electrical Engineeringand Computer Science, synthesizedLPC signals, and error prediction signals from the LPC. Re- MassachusettsInstitute of Technology,Cambridge, MA 02139) sultsindicated high levels of performancein the natural and error predic- Becauseof acousticsimilarities between some letters of the alphabet, tionsignal conditions and performance near chance in thenoise excited automaticrecognition of continuouslyspoken letters is a difficulttask. condition,thus suggesting that the error signal provides valuable informa- Thegoal of thisstudy is to determineand compare how well listeners and tionthat allows humans to distinguishbetween speakers. It maybe possi- spectrogramreaders can recognizecontinuously spoken letter strings bleto improveverification algorithms by adapting current models to more from multiplespeakers. The interestin spectrogramreading results is accuratelyutilize information used by humanlisteners.
S55 J.Acoust. Soc. Am. Suppl. 1, Vol. 83, Spring 1988 115thMeeting: Acoustical Society of America S55 WEDNESDAY AFTERNOON, 18 MAY 1988 GRAND BALLROOM B, 1:00 TO 2:30 P.M.
SessionZ. Physiological Acoustics II: Auditory-Evoked Responses
John H. Mills, Chairman Medical Universityof SouthCarolina, Department of Otolaryngologyand CommunicationSciences, 171 Ashley Avenue, Charleston, South Carolina 29425
ContributedPapers
1:00 The existenceof corticalasymmetries in humansis well documented andparalleling this is thesuggestion that speechsounds are processed by Zl. Age-relatedchanges in auditorybrain-stem responses of Mongolian differentcentral neural pathways than nonspeech sounds. This study pre- gerbil. John H. Mills, Richard A. Schmiedt,Joe C. Adams, Bradley sentsdata that reveallower brain-stem asymmetry to a low-frequency A. Schulte,and Lawrence R. Kulish(DepartmentOtolaryngologyand sinusoidal.Frequency following responses (FFR) (500 Hz) were ob- Communication Sciences,Medical University of South Carolina, tainedfrom ten normal hearing subjects monaurally and binaurally over a Charleston, SC 29425) rangeof intensities.The resultsprovide clear evidence that the mortaural As part of a largestudy of age-relatedhearing loss, the Mongolian FFRs have differencesin onset latency and differencesin latencies gerbilas an animal model is evaluated. Part of thiseffort involves monitor- throughoutthe response waveform as well as diversities in theirampli- ing electricalpotentials recorded from the auditorynerve and brainstem. tudes. Since the FFR emanates from sites that are not rostral to the inferi- The gerbilsare born and raised in an acousticallytreated quarters where or colliculus,the findings support the viewthat the lowerauditory brain the mediansound level is 35 dBA. Testsignals are 1.8-mstone pips rang- stemis functionallyasymmetrical. ingfrom 1-16 kHz in octavesteps. By age22-24 months,auditory sensi- tivity is decreasedby about10 dB at mostfrequencies although there are large differencesbetween animals. By age 36 months,mean threshold changesare largestat 8 and 16 kHz, about35 dB, and decreaseto 10-20 1:45 dB at lowerfrequencies. The meandata for the36-month group form an audiometricconfiguration that isqualitatively similar to that observedfor ZA. The effect of stimulusrepetition rate on interaural ABR latency 60. to 65-year-oldhuman males and 70-year-old human females. Individ- differencesin normaland hearing-impaired subjects. Maureen Hannley, ualdifferences are large with some animals having nearly normal hearing Lorene Wcichert,and SarahB. Wedekin(Division of Otolaryngology and othershaving hearing losses of 50-70 dB at all test frequencies. and Head/Neck Surgery,Stanford University Medical Center, Stanford, [Work supportedby NINCDS PO NS-25039.] CA 94305) Twenty-fivesubjects with normalhearing and 15 with bilateralhigh- frequencysensorineural hearing impairment were presented click stimuli at four repetitionrates ( 11, 33, 67, and 89 clicks/s) to accessinteraural 1:15 symmetryof waveV latencyand amplitude. Predictably, amplitude was significantlydecreased and latencyincreased across all subjectsat the Z2. Auditory responsesto the envelol•sof pseudorandomnoise stimuli in higberrepetition rates. In bothgroups, however, interaural latency differ- humans.Robert A. Dobie and Michael J. Wilson (Department of ences(ILDs) remainedwith the 0.3-msnormal criterion previously es- Otolaryngnlogy,RL-30, Universityof Washington,Seattle, WA 98195) tablishedfor slowrepetition rates. Thus, both abnormal repetition rate Averagedscalp potentials evoked by continuouspseudorandom noise effectsand interaurallatency differences can serveas reliable indicants of canbe crosscorrelated with the evokingstimulus, yielding a cross-corre- retrocochlcarauditory dysfunction, even when peripheralhearing im- lationfunction (CCF) that reflectsneural phase locking and is quite sen- pairmentexists. sitivefor low-frequencystimulus components [M. J. Wilsonand R. A. Dobie, Electroencephalogr.Clin. Neurophysiol.66, 529-538 (1987)]. However,for higherfrequency signals, replicable CCFs canonly be ob- tainedat moderateto highintensity. Since auditory neurons also respond 2:00 to envelopesof complexsounds, even for high-frequencycarriers, scalp responseswere comparedto the envelopesof complexsounds; the resul- ZS. The auditory brain-stemresponse in children with a history of OME. tant envelopecross-correlation functions (ECCFs) containedreplicable Ron D. Chambers,Lynne E. Rowan (Department of Speechand responsecomponents primarily below 1000 Hz, regardlessof theevoking Hearing Science, University of Illinois, 901 South Sixth Street, stimulusspectrum. The ECFF thresholdsfor three octave-bandstimuli Champaign, IL 61820), and Michael A. Novak (Carle Clinic (830-1562, 1611-3125, and 3174-6201 Hz) were more sensitive than Association-4W,602 West University,Urbana, IL 61801) CCF thresholds(p = 0.018), averaging35-dB spectrum level for tennor- mal subjects.When stimuliwith only odd harmonicswere used, replicable The auditory brain-stemresponse (ABR) was usedto evaluatethe odd-componentresponses were seen only in thespectral range of thestim- auditory-neuralstatus of childrenwith a historyof frequentor persistent uli, while even-componentresponses (presumably to stimulusenvelope ) otitis media with effusion(OME). Subjectswere selectedfrom the case- were seenonly in the low-passedscalp responses. loadof a boardcertified otolaryngologist and from the populationat large. They were pair-matchedby age and sex to form an otitis-proneand a controlgroup. The otitis-pronegroup included only thosechildren whose degreeof OME involvementwas sufficientto warrant tympanostomy tubeplacement. Stimuli wererarefaction clicks presented at 85 dB nHl at 1:30 a rate of 21.4/s. The latency of wave Ill and the Ill-I interwaveinterval were longerfor the otitis-pronesubjects than for the control subjects, Z3. Human auditory brain-stemusymmetry in the frequencyfollowing which agreeswith earlier results[R. C. Folsom,B. A. Weber, and G. response(FFR). BallachandaB. Bopannaand GeorgeMoushegian Thompson,Ann. Otol. Rhinol-Laryngol.92, 249-253 (1983)]. The re- (Callier Center for Communication Disorders, University of Texas at sultsand precautionsin interpretationwill be presented.[Work support- Dallas, 1966 Inwood Road, Dallas, TX 75235} ed by Carle ResearchFoundation. ]
$56 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America $56 2:15 wereanalyzed. The latenciesof wavesI, V, andthe I-V interpeakinterval wereexamined as a functionof subjectage, subject gender, operative site, Z6. lntrao•rative variability of auditorybrain-stem responses during andoperative stage. The resultsindicated that: ( 1) subjectage was not a cranial nerve microvascular decompression.Maureen Hannley significantfactor in intraoperative ABR variability;(2) significantmale- (Division of Otolaryngology,Stanford University Medical Center, female latency differenceswere presentonly in the earlier operative Stanford,CA 94305), Set-Mix Huang (Departmentof Physiology, stages;{ 3) greaterlatency changes were produced during decompression LouisianaState University Medical School,New Orleans,LA 70112), of the facialnerve than during trigeminal nerve decompression; and (4) and PeterA. Randzeus(Barrow NeurologicalInstitute, Phoenix, AZ greaterresponse variability was present during the later operative stages. 85001 ) The patients'postoperative auditory status was studied to determinethe The auditorybrain-stem response (ABR) recordingsof 40 patients predictiveefficiency of threecurrently used criteria for significantintra- who underwentmicrovascular decompression of cranialnerves V and VII operativelatency change.
WEDNESDAY AFTERNOON, 18 MAY 1988 WEST BALLROOM A, 1:20 TO 2:50 P.M.
SessionAA. PhysicalAcoustics IV: MiscellaneousTopics in PhysicalAcoustics
Steven R. Baker, Chairman PhysicsDepartment. Code 61B•4, Naval PostgraduateSchool, Monterey, California 93943
ContributedPapers
1:20 1:50
AAI. Evolution of unstable disturbances at a mean-flow stagnation AA3. Photoacousticmethod for measuring high-order multiphoton point. CharlesThompson and Martin ManIcy (Department of absorptionin transparentmaterials. Scott C. Jones(Department of Electrical Engineering,University of Lowell, I University Avenue, Physics,Washington State University, Pullman, WA 99164) Lowell, MA 01854) A photoacoustictechnique for measuringmultiphoton absorption The interaction between infrasonic, acoustic disturbance and the fromsharply focused pulsed laser beams is developed for arbitraryorder mcanflow near the stagnationpoint of a bluffbody will beexamined. The of nonlinearabsorption. The resultingsound source is neitherspherical criterionfor stabilityand intermodulation of thedisturbances will bedis- norcylindrical, and a multipoleexpansion is presentedthrough quadru- cussed.The spatialand temporal evolution will alsobe addressed. It will poleterms for thermaland electrostrictive source terms. The analysis be shownthat infrasonicand unstablevortical disturbances, working in showsthat theacoustic amplitude is proportionalto absorbedenergy, and conjunction,can serveto modulatethe amplitudeof the acousticwave. proportionaltothe ruth power of incident laser energy under conditions of For a harmonicallytime-varying acoustic wave, this modulation is mani- m-photonabsorption. The calibratedmeasurement of the four-photon festedin an increasein the spectralbandwidth of the wave. [Work sup- absorptioncross section of NaCI at wavelength532 nm is presented. portedby AnalogDevices Professorship.] [Work supportedby NSF.]
2.'O5
1:35 AA4. Acoustic imaging by a method of image projections. Daniel Rouseif,Robert P. Porter, and Leung Tsang(Deparlment of AA2. Effect of internal relaxation on the optnacousticsignal in CS2. ElectricalEngineering, FT-10, Universityof Washington,Seattle, WA CharlesH. Thompson,Henry E. Bass,and Richard Raspet(Physical 98125) AcousticsResearch Laboratory, University of Mississippi,University, A methodof three-dimensionalacoustic imaging based on imagepro- MS 38677) jectionsis derived.A weaklyscattering object is illuminatedby a plane Theoptoaeoustie signal in CS2has been observed using an experimen- wave and the scattered field is recorded. Inversion methods based on tal eonfigurationsimilar to thatof Sullivanand Tam [B. Sullivanand A. projectionsthat operate directly on thescattered field are inaccurate due C. Tam, J. Acoust.Soc. •-n. 75, 437-441 (1984) ] and hasbeen compared to diffractioneffects. By backpropagatingthe scattered field onto a plane to theoryand similar results in propanol.The comparisons illustrate that in theimage region, diffraction effects are reduced and a projectionof the an internalrelaxation process is affectingthe observedwaveform. Assum- objectis generated. Conventional x-ray inversion techniques may then be inga simplethree-level system for internalenergy and using independent- usedto imagethe object.The relationshipbetween the backpropagated ly measuredvibration-to-translation transfer rates leads to theconclusion fieldand the projection is derived both for high-frequencyincident waves thatthe relaxation process observed is associated with electronic-to-trans- and low spatialfrequency scatterers. Backpropagation additionally al- lationor electronic-to-vibrationenergy transfer with a relaxationtime of lowsthe use of curvedor misalignedrecording surfaces. [ Work supported 90 ns. [ Worksupported by theOffice of NavalResearch. ] by ONR. ]
S57 J.Acoust. Soc. Am. Suppl. 1, Vol. 83, Spring 1988 115thMeeUng: Acoustical Society of America S57 2:20 peraturedependence from previously reported values. [Work supported by ONR. ] AA$.Propagation of soundin vibrationallyexcited N2/He, N•/H2, N2/ CH4,and N2/H:O mixtures.F. DouglasShields and L. DwynnLafieur (NationalCenter for PhysicalAcoustics, P.O. Box847, University, MS 2:35 38677) AA6. Instability and particle agglomerationin acousticallydriven flows. CharlesThompson and Vineet Mehta (Department of Electrical Engineering,Laboratory for AdvancedComputation, University of Previousmeasurements of resonantreverberation of soundin N2/H 2 Lowell, 1 UniversityAvenue, Lowell, MA 01854) mixturesin a closedtube have shown an amplification of thesound follow- ingthe rapid excitation of thegas by an electric discharge [J. Acoust. Soc. The mechanismresponsible for instabilityand turbulencegenerated Am. gl, 87 (1987)]. Thesemeasurements have been extended in this by highamplitude acoustic waves has been of interestof in acousticagglo- paperto includeN2/He, N2/CH4,and N,_/H:O mixtures. Some of the merationof particles.When a low-frequencyoscillation is usedin con- excessivelylarge amplification previously observed in N:/H2 mixtures junctionwith a higherfrequency acoustic wave, localized unstable bound- has now been attributed.to nonlinearity in the microphoneresponse. ary layerdisturbances can be used to drivethe outer flow field. By virtueof However,even after correcting this error the amplification is muchlarger the Reynoldsstress, the harmonicspectrum of the unstablevortical dis- andlasts for a longertime than theoretically predicted. The variationin turbanceserves to modulatethe acousticwave amplitude. This process the translationaltemperature following the dischargeis monitoredby enhancesthe ability of theacoustic wave to agglomerateparticles. A theo- measuringthe sound velocity. Relaxation times determined from the tem- reticalmodel of theinstability mechanism and the interactionprocess will peratureversus time curves differ only slightly in magnitudeand/or tern- be presented.[Work supportedby AnalogDevices Professorship.]
WEDNESDAY AFTERNOON, 18 MAY 1988 ASPEN ROOM 1:30 TO 2:30 P.M.
SessionBB. PsychologicalAcoustics IV and Educationin AcousticsIII: Auditory Demonstrationson Compact Disc
W. Dixon Ward, Chairman HearingResearch Laboratory, University of Minnesota,2630 UniversityAvenue, $.E., Minneapolis,Minnesota 55414
Invited Paper
1:30
BBl. Auditorydemonstrations on compactdisc. A. J. M. Houtsma(Institute for PerceptionResearch, P.O. Box513, 5600 MB, Eindhoven,The Netherlands),T.D. Rossirig(Department of Physics,Northern Illinois University,DeKalb, IL 60115),and W. M. Wagenaars(Institute for PerceptionResearch, P.O. Box 513, 5600 MB, Eindhoven,The Netherlands)
In 1984,the Committeeon Educationin Acousticsof the ASA discussedthe possibilityof reissuingthe "HarvardTapes," a seriesof audiodemonstrations ontape cassettes originally published in 1978by DavidM. Greenat the Harvard Laboratoryof Psychophysics.The presentauthors undertook the job of evaluatingand updatingthe content of theoriginal tapes, producing new audio material, and writing a newexplanatory book. Thecompact disc (CD) mediumwas chosen to replacemagnetic tape. Through substantial cooperation from EindhovenUniversity of Technology,Northern Illinois University, the PhilipsCompany, and the ASA, a CD witha 92-pagebooklet was developed containing groups of auditory demonstrations divided over seven general topics.For eachgroup, a generalintroduction with basicreferences is providedin the booklet.For each demonstrationwithin a groupa detailedbackground, explanation and further literature references are given. Verbalintroductions on the CD havebeen kept very brief and serve mainly to identifyeach demo. Demonstra- tionsor partsthereof can be calledindividually in anyorder, or preprngrammedin arbitrarygroups, through 80 separatetrack or "title"numbers. One thousand copies of theCD sethave been made on a firstproduction run. They are availablethrough the headquartersof the ASA. A CD playerwith the auditorydemonstration will be availableat the registrationdesk for the entire meeting.
S58 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115th Meeting:Acoustical Society of America $58 WEDNESDAY AFTERNOON, 18 MAY 1988 WEST BALLROOM B, 1:30 TO 2:50 P.M.
SessionCC. Underwater AcousticsIV: Object Scattering
StephenA. Reynolds,Chairman AppliedPhysics Laboratory, University of Washington,1013 N. E. 40thStreet, Seattle, Washington 98105
Chairman's Introduction--l:30
ContributedPapers
1.'35 2:05
CCI. Pulsescattering from an objectsubmerged in the ocean.Michael CC3. Harmonic angularperturbation of a toroidal wavefront:A simple D. Collins and Michael F. Werby (NORDA Numerical Modeling unfoldingof an axial caustic.W. Patrick Arnott and Philip L. Marston Division, NSTL, MS 39529) (Departmentof Physics,Washington State University, Pullman, WA 99164-2814) The low grazingangle asymptotic limit, whichis usedto derivethe parabolicequation, is usedto derivean efficientcomputational method Axial causticscan be observedin backwardand forwardscattering for propagationand scattering of acousticpulses in theocean. In orderto fromhighly symmetric surfaces such as spheres. A wavefrontthat is local- avoidthe time-consumingfrequency decomposition, all calculationsin- ly toroidal,propagates to producean axialcamtic. The unfoldingof the volring the acousticfield are donein the time domain.The time domain axialcaustic generated by a harmonicperturbation of thewavefront shape equivalentof theparabolic equation lB. E. McDonaldand W. A. Kuper- in the aximuthaldirection will be considered.The perturbedwavefront man,J. Acoust.Soc. Am. 81, 1406-1417(1987) ] is usedfor propagation shapeWis given by W(s,•) = [ (s -- b}•/2ct] + A cos(2•), in polarco- of theincident field up to thescatterer and for propagationof thescattered ordinates(s,•), wherea and b specifythe characteristicradii of the un- fieldaway from the scatterer.Waterman's extended boundary condition, perturbedtorus and A isa measureof theperturbation. The caustic asso- whichis a frequencydomain method, is usedto constructa time domain ciatedwith W consistsof four connectedtransverse cusps in a distorted integraloperator that transforms incident waveforms into scattered wave- diamondshape known as an asteroid.The causticshape has been identi- formsand depends only on thegeometry and composition of thescatterer. fiedfor observationpoints from the nearzone to the far zone.Locations Like theparabolic equation method, this method offers an attractivecom- on W that contributerays near a cusppoint, have been identified analyti- binationof accuracyand efficiency.[Work supportedby ONR and cally.Observations of the asteroid caustic have been made by scatteririS NORDA. ] lightfrom freely rising oblate gas bubbles in water.The bubbles are oblate sincethey are larger than those considered in ourprevious study of back- scattering[W. P. Arnottand P. L. Marston,1. Opt. Soc.Am. A (in press)]. Thefar zone scattering exhibits an asteroid caustic decorated by a diffractionpattern. The calculated caustic agrees favorably with observa- tions. This calculation should also be useful in acoustic reflection and scatteringproblems. [Work supportedby ONR.] 1:50
CC2. Principal curvaturesof general wavefrontsand of reflecting or refracting surfaces. Cleon E. Dean, W. Patrick Arnott, and P. L. Marston (Department of Physics,Washington State University, Pullman, WA 99164-2814)
A nonparametricderivation of the exactform of the two principal curvaturesg•.2 of generalwavefronts and reflecting or refractingsurfaces 2:20 was madefor Carlesianand polar coordinates.The standardexpression for the Gaussiancurvature K•K2 was recovered in the Cartesiancase. For CC4. Comparisonof nsymptotieand exact resultsfor scatteringfrom the polarcase let rigid spheroids.Michael F. Werby and Michael D. Collins (NORDA NumericalModeling Division, NSTL, MS 39529) H----1 , W•W, I 2 W,oWo_ Two,I Resultsgenerated with theKirchhoff approximation, which is effec- B----I W•+ I W•W• 2 2 tiveFor backscattering, and the diffraction integral method, which is effec- tivefor forwardscattering, will becompared with theexact solution ob- tained with Waterman's extended boundary condition method. I +__1 I Calculationswill be presentedfor kL/2 rangingfrom 5 to 300 andfor le aspectratios ranging from I to 15.An additionalasymptotic approxima- andC- I -F W•, + (l/P) W•, wherez = W(r,O)is thepolar equation tion, the methodof stationaryphase, is usedfor higherfrequencies. For of thesurface and W• = •W'/•r, W•, = • • W/•?, andso fo•h; theanal- higheraspect ratio targets, one would expect the accuracyof high-fre- ysisgives •.• = [B +_( B • --4HC}'n]/(•C•n). TheGaussian curva- quencymethods to be difficult to predictand to dependnot only on kL/2 tureis • • H [C • • and• reduceto previous results [ D. G. Burkhard hutalso on bistaticconfigurations due to the largevariation of Gaussian and D. L. Shealy,Appl. Opt. 20, g97-909 (1981)] in the specialcase curvatureon thesetargets. Hence, the focusof thepresentation will beto • = W(r). Theseresults should bc usefulfor generalizedray tracingand illustratehow accuracy depends upon kL/2, angleof incidence,and angle for locatingcaustics in catastropheacoustics and scattering calculations. of observationfor high aspectratio taxgets.[Work supportedby ONR [Work supportedhy ONR.] and NORDA.]
S{i9 J. Acoust.Soc. Am. Suppl. 1. Vol.83. Spring1988 115thMeeting: Acoustical Society o! America 2:35 the incidentfield to the scatteredfield in a waveguide.The transition matrixis obtained from the extended boundary condition (EBC) method. CC5. The effectof an objectin a waveguideon the field producedby a This solutionis coupledwith the waveguidesolution [G. V. Norton and point source.Guy V. Norton and M. F. Werby (Naval OceanResearch M. F. Werby,"Some Numerical Approaches to DescribeAcoustical Scat~ and DevelopmentActivity, NSTL, MS 39529-5004) tering from Objectsin a Waveguide,"in Proc. Sixth Int. Conf. Math. Modeling,St. Louis,MO, 4-7 Aug. 1987].This method satisfies all ap- Whenthe acoustic wave produced by a pointsource in a shallowwater propriateboundary conditions and yields a continuoussolution through- waveguideensonifies an object,the objectacts as a secondarysource and outthe waveguide. The objectused is a rigidspheroid of aspectratio 5:1. in turngenerates a guided wave. The effect that the object has on the initial The frequenciesare 100 and 450 Hz. The source,receiver, and object incidentfield is estimated.In particular,comparisons are made with depthsare differentcombinations of 50, 75, and 200 m, and two wave- transmissionloss versus range for the point source by itself and with trans- guidewater depthsat 150 and 400 m are considered.The natureof the missionloss versus range for theobject as ensonified by thepoint source. effecton theinitial incident field due to theobject and as influenced by the To obtainthe scatteredfield from the object,the scatteredfield is first relativeplacement of the sourceand receiverrelative to it is demonstrated. generatedin thevicinity of theobject using a transitionmatrix that relates [Work supportedby NORDA. ]
WEDNESDAY AFTERNOON, 18 MAY 1988 EAST BALLROOM B, 1:45 TO 3:00 P.M.
SessionDD. ArchitecturalAcoustics III: Topicsin General ArchitecturalAcoustics
Roy L. Richards, Chairman Towne,Richards & Chaudiere,Inc., 105N.E. 56thStreet, Seattle, Washington 98105
ContributedPapers
1:45 thickness,density, etc. ) andthe effect of providingfor airflowhave been examined.Laboratory results and preliminary field test data will bepre~ DDI. Meusuringsound transmission through suspended ceilings. J. sented. D. Quirt and R. E. Halliwell (AcousticsSection, Institute for Research in Construction,National ResearchCouncil of Canada,Ottawa KIA 0R6, Canada) In mostmodern office buildings, interoffice partitions stop at thesus- pendedceiling. Sound insulation between such offices is generallyinade- 2:15 quatefor speech privacy. This paper presents the first stages of a projectto developpractical acoustical solutions. A newlaboratory has been con- DD3. Test resultsfor floorand ceiling modifications for impactisolation. structed;this simulates a pairof adjacentoffices with a commonplenum EdwardA. Daly (Daly EngineeringCompany, 11855 S. W. Ridgecrest spaceabove the suspended ceiling. Characteristics of thetest environment Drive, Beaverton, OR 97005) and Kerrie G. Standlee (Van Gulik/ and pertinentstandard test procedureshave beenstudied. A seriesof Oliver, Inc., 543 Third Street,Lake Oswego,OR 97034) suspendedceiling systems has been tested. The resultsindicate the relative The impactsolution evaluation methods presented by thetwo authors importanceof majorphysical factors controlling sound transmission in a papergiven at the 112thMeeting ofASA [J. Acoust.Soe. Am. Suppl. throughtypical suspended ceilings. I 80, S56 (1986) ] wereused to evaluatemodifications of floor/ceiling systems.Tests were conducted on thepresently most-used floor structure, the wood/concretestructure. Some of the floormodifications produced significantimpact soundisolation improvements. However, the floor modificationsthat produced the best sound attenuations provided unsta- ble flooringconditions. All ceilingmodifications produced significant im- pactsound isolation improvements. The typeof constructionrequired for 2:00 all ceilingmodifications were found to be acceptableto the construction trades.An evaluationmethod developed by oneof the authorswas usedto DD2. Fuzzwall:Blocking sound transmission above suspended ceilings. J. producea singlenumber effectiveness rating for the floor/ceilingmodifi- D. Quirt and R. E. Halliwell (AcousticsSection, Institute for Research cations.The resultsof thesetests indicate ceiling modifications are more in Construction, National Research Council of Canada, Ottawa K1A effectivethan floormodifications in providingimpact isolation. 0R6, Canada)
Interofficepartitions commonly stop at the suspendedceiling. The plenumspace above the ceilingis usedas an air-returnfor the ventilation 2:30 system.Sound transmission through this spaceoften limits interoffice speechprivacy. This paper reports the performance of a barriersystem DIM. The measurementof acousticreflection coefficients by using intendedto reducesound transmission without preventing aidlow. Fi- cel•trul techulqu•s.J. Stuart Bolton (Ray W. Herrick Laboratories, brousabsorptive material has been used as a barriersystem. Dependence Schoolof MechanicalEngineering, Purdue University, West Lafayette, of the noisereduction on physicalparameters (flow resistance,barrier IN 47907)
S•0 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115th Meeting:Acoustical Society of America S60 The authorh•s recentlydeveloped a new free field technique for thein 2.•5 situ measurementof acousticreflection coefficients by the useof transient acousticsignals [ I. S. Boltonand F• Gold, J. Soundrib. 110(2), 179-202 DDS. A boundaryintegral equationmethod for transient problemsin (1986)]. The measurementis conductedby usingan omni-directional architecturalacoustics. D. K. Holger (Departmentof Engineering loudspeakerto generatea testsignal; a microphoneplaced between the Scienceand Mechanics,2019 BlackEngineering Building, Iowa State loudspeakerand the reflectingsurface is usedto detectboth the incident University,Ames, IA 50011) (i.e., the direct) and reflectedsignals. Cepstral processing was used to eliminatesome of thedifficulties associated with existingreflection mea- A numericalmethod for solvingtransient interior acoustic field prob- surementmethods. Specifically, in the newtechnique the directand re- lemsby combiningboundary integral equation and inverse Fourier trans- fleetedsignals are measuredsimultaneously and are allowed to overlapin form methodsis described.An indirectboundary integral equation for- time,thus relaxing geometrical constraints. Further, if theincident signal mulation of the Helmboltz equationusing hybrid layer potentialsis is sufficientlybroadband, a knowledgeof its spectrumis not requirednor combined with a numerical inverse Fourier transform to obtain time do- is it necessaryto measureit individually.The techniqueis basedon the mainsolutions for acousticfield variables. Frequency domain representa- directextraction of theimpulse response of thereflecting surface from the tion of transientsources, frequency domain windowing of the integral powercepstrum of a signalcontaining both the directand reflectedsig- equationsolutions, and frequencydomain resolution enhancement are nals. Full-scaleacoustical trials of this techniquewill be describedand discussed.Numerical results are presentedfor the acousticfield inside a measurementsof the reflectioncoefficients of samplesof polyurethane n etangularprism with various boundary conditions and for various tran- foam will be shown. sient source excitations.
WEDNESDAY AFTERNOON, 18 MAY 1988 GRAND BALLROOM C, 3:00 TO 5:00 P.M.
Plenary Session
ChesterM. McKinney, Chairman
President,Acoustical S•ciety of America
BusinessMeeting
ProposedAmendment of the Bylawsof the Society
Presentation of Awards R. Bruce Lindsay Award to Gilles A. Daigle Gold Medal to Arthur H. Benade Gold Medal to Richard K. Cook
Invited Speaker The HonorableGuy Strutt will talk about his grandfather,Lord Rayleigh.
Musical Demonstration StuDempster of theUniversity of WashingtonMusic Department will demonstratea didjeridu, an aboriginal Australian musical instrument.
S61 J. Acoust.So(=. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America THURSDAY MORNING, 19 MAY 1988 EAST BALLROOM B, 8:00TO 11:45A.M.
SessionEE. EngineeringAcoustics IV: Laboratoryand Measurement Condenser Microphones
GeorgeS. K. Wong,Chairman NationalResearch Council of Canada,Physics Division, Montreal Road, Ottawa, Ontario K1J OR6,Canada
Chairman's Introduction.--8:00
Invited Papers
8:05
EEl. Electretcondenser microphones in physicalacoustics research. S. L. Garrett (PhysicsDepartment, Naval PostgraduateSchool, Monterey, CA 93943) The simplestcondenser microphone in termsof constructionand use is a charged,aluminized Teflon membraneplaced directly over a conductingback plate. In thisconfiguration, the low-frequency open circuit sensitivityisdetermined approximately by the product of the"effective" electret bias voltage and the compress- ibilityof thefluid trapped between the membrane and the backplate, typically 1 mV/Pa in air.This talk will describethe procedures for calibrationof sucha transducerin a resonatorand its application to themeasure- mentof constituents in binary or pseudobinarygas mixtures. Other aplications to low-temperaturephysics will be describedthat includeacoustic measurement of temperaturefluctuations with a perforatedclectret mem- branehaving a detectionthreshold of = 10nanokclvin/v/Hz in superfiuid4He at 1 dcgabove absolute zero, andthe measurement ofthe temperature dependence ofthe order parameter ofsuperfluid 3He-A and 3He-B at temperaturesofr• of a degreeabove absolute zero. [ Worksupported by the Office of NavalResearch. ]
8:25
EE2. Laboratorycalibration of standardand measurementcondenser microphones. Victor Ncdzclnitsky (NationalBureau of Standards,Building 233 ( Sound), RoomA 149,Gaithersburg, MD 20899) Thereciprocity technique isthe most highly developed and most widely used primary calibration method at themajor national standards laboratories of the world. Critical measurements of sound pressure in air are usuallytraceable to suchcalibrations, which are briefly described with referenceto specificapparatus and proceduresat the NBS.Secondary methods, including those based upon reciprocity, can offer improved conveniencewith minor to significantsacrifices in accuracy and precision. Dynamic signal analyzers incorpor- atingdigital signal processing are now sufficiently accurate and precise to offerpromise for improved and/or morereadily automated measurements, especially secondary free-field calibrations. Specific examples from the NBS research and measurement services are discussed.
8:45
EE3. Microphonecomparison with the three-porttwo-microphone cavity. George $. K. Wongand Tony F. W. Embleton(Division of Physics,National Research Council Canada, Ottawa, Ontario K 1A0R6, Canada) Theperformance of the three-port two-microphone cavity [Wong and Embleton, J. Acoust.Soc. Am 71, 1276-I277 (1982) ] forthe comparison ofcondenser microphones isdescribed. The experiment involves, 1-, «-, andl-in. microphones from three manufacturers. With a referencemicrophone, the above cavity is able to measure,by comparison, the sensitivities and frequency response of varioussizes of microphonesunder differ- entphysical conditions, such as with or withoutprotective grids. For frequencyresponse measurements over thefrequency range from 20 Hz to severalkilohertz, the experimental results are similar to thoseobtained with the electro-actuatormethod, and the cavitymeasurement does not requireany access to the air spacethat is in closeproximity to themicrophone diaphragm. In general,the frequency response performance of the cavity improveswith the smallersize microphones.
9:05
EFA.Scattering of soundby finite cylinder•A preliminarystudy applicable to microphones.A. F. Sey•ert, M.L. Lin,and T. W. Wu (Departmentof MechanicalEngineering, University of Kentucky,Lexington, KY 40506-0046) Whenthe wavelength ofsound is of the same order as the diameter of a microphone,the diffraction ofsonnd in thenearfield of themicrophone will alter the indicated sound-pressure amplitude. This fact has long been
S62 J.Acoust. Soc. Am. Suppl. 1, Vol. 83, Spring 1988 115thMeeting: Acoustical Society of America S62 observedexperimentally, and the effect of diffractionis consideredin thedesign of microphones.However, whena pairof microphonesisused, as with sound intensity measurements, diffraction alters the amplitude and phasedifference indicated by themicrophones, and these effects become important at frequencieswithin the normaloperating frequency range of a singlemicrophone of thesame diameter. It hasbeen difficult to predict diffractioneffects from analyticalmodels, due to the complexgeometry involved. A numericalmethod has beendeveloped, based on the boundaryelement method, to studythe scatteringof soundby obstaclesof arbitraryshape. As a preliminarymodel of a microphone,the finite cylinder was used. In thispaper numerical resultsare presented for scattering from a singlefinite cylinder and from a a pairof cylinders. [ Work supported in part by the NSF.1
9:2.,-e.-9:40
Break
EES. Intensity measurementmicrophones and their calibration. Erling Frederiksen(Bruel & Kjaer Development,Naerum Hovedgade 18, DK-2850, Naerum,Denmark) New microphonesespecially designed for intensitymeasurements have madeit possibleto make valid intensitymeasurements under soundfield conditionsthat previouslywould lead to invalid results.The achievedimprovements are mostimportant at low frequenciesbu! mayalso be significant, for instance,at 500 Hz, dependingon thesound field. The microphonesof thisnew type are partly characterized by a stabilityand spreadin their low-frequencyphase characteristics that are a factor of 10-15 timesbetter and partly by a sensitivityto soundpressure at the staticpressure equalization vent that is severalorders of magnitudelower thanthat of conventionalmicrophones. These properties are all importantfor the improvementof intensity instrumentsthat arebased on microphonesoperating on thepressure principle, as the phasedifference between exactlyspaced points in soundfields need to bemeasured with great accuracy. System phase errors of theorder 0.01 degare of interestand seem to becomepossible by propercalibration and correction. The useof conven- tional microphonesimplies significant erors due to pressuregradients errors that are eliminatedby the ex- tremelylow vent sensitivityof the new microphones.Most leadinglaboratories have spent great efforts on magnitudecalibration of microphonesensitivity but verylittle on phasecalibration. The microphonesand the methodsused for verificationof microphoneand probe properties are described in thispaper.
ContributedPapers
10:00 configurationis mostcommonly used for the microphoneprobe. With theseconventional configurations, however, two tubescontaining indi- EE6. Study on the linearity of laboratory standard condenser vidualpreamplifiers are paralleland closeto eachother and causea dif- microphone. S. Yoshikawa (Kanagawa Institute of Technology, fraction problem.A recentlydeveloped one-dimensional four-micro- Kanagawa,243-02 Japan) phoneprobe is structurely much simpler than conventional probes. Two Becauseof themany features such as wide frequency range, high sta- microphonepairs cover low- (50-800 Hz) andhigh- (800-10 kHz) fre- bility,good transient response, etc., the condensermicrophone has been quencyranges and, therefore, it is possibleto measurethe whole frequen- usedas a laboratorystandard microphone and a measurementmicro- cy rangewithout changing the microphonedistance. The effectivedis- phone.One problemis that the condensermicrophone has essentially tancebetween pair microphones,pressure, particle velocity, and active nonlinearcharacteristics on the electricaloutput/acoustical input func- intensityresponses are measuredin an anechoicchamber. Assuming a tion.As sound-pressurelevel becomes higher, nonlinearity becomes larg- farfieldcondition, the measurement accuracies of pressure,particle veloc- er. In this study,linearity of the laboratorystandard condenser micro- ity, andintensity responses are checked by comparing these with thepres- phoneat high sound-pressore level was investigated. In thecase of the«-in. sureof a referencemicrophone. The accuracyof the reactiveintensity condensermicrophone MR- i 12,linearity was ensured up to 120dB SPL. measurementis checkedby measuringthe pressuresquare gradient by a Over 120dB SPL, output/inputfunction becomes larger and distortion referencemicrophone. The resultswill be shownat the meeting. becomesdetectable. By experiment, measured distortion factor was 2% at 151 dB SPL and 5% at 154 dB SPL. The nonlinearityat high sound- pressurelevel was studied analytically, and the effectof someparameters suchas shape of backelectrode, biasing voltage, and stray capacitance was 10:30 clarified.Based on the analyticalstudy, the possibilityof loweringthe nonlinearitywas investigated. A methodto realizethe condensermicro- EES. Mounting and selection of microphones for two-microphone phonehaving low distortionand highsensitivity was suggested. impedance measurements.Gordon Ebbitt and J•rgen Christensen (Bruel & Kjaer, NaerumHovedgade 18, DK-2850 Naerum,Denmark) The determinationof acousticimpedance as specifiedin ASTM E- 10:15 1050requires the measurement of soundpressure at twopositions along a EE7. Measurement aeeueaeyof a one-dimensionalfour-microphone standingwave tube. The choiceof microphonesand their mountingcan sound intensit• probe. Hidao Suzuki, • Anzai, and have a significantinfluence on the measurementresults. A variety of mi- Takahiko Oho (Oho SokkiCo., Ltd., 2-4-1 NishishinjukuShinjuku-ku, crophonetypes (i.e., phase-matchedcondensers, probe tube types,con- Tokyo 163,Japan) densermicrophones fitted with phasecorrectors, etc. ) will beconsidered and the resultsfrom thesevarious types will be compared.The mounting In thesound intensity measurement, it is important that a microphone of the microphones(i.e., flushwith the tubewalls, near the middleof the probedoes not disturbthe soundfield. A side-by-sideor a face-to-face tube, etc. ) will also be considered.
S63 J. Acoust.Soc. Am. Suppl.1, VoL 83, Spdng1988 115th Meeting:Acoustical Society of America S63 10:45 Thepresent paper initiates astudy to understand the errors inherent to a laboratorytechnique similar to that usedby K. J. Taylor [J. Acoust. EE9. Field calibration and verification of sound intensity systems. Soc.Am. 70, 939-945 ( 1981) ] for absolutecalibration of microphones. Gordon Ebbitt and Erling Frederiksen (Bruel & Kjaer, Naerum Consideredhere is the idealizedsituation when a plane-polarizedmono- Hovedgadc18, DK-2850 Naerum, Denmark) chromaticplane electromagnetic wave is incidenton a finitescattering A two-microphonesound intensity system is, in general,calibrated by volumecontaining dispersed small dielectric spheres that oscillateunder usinga pistonphoneor a soundlevel calibrator to set the sensitivities ofthe thejoint influenceof a planetransient acoustic wave and Brownianmo- two measurementchannels. While this is all that is requiredfor a calibra- tion. Any componentof the scatteredand Doppler-shiftedelectromagnet- tion of the system,it doesnot provideany checkof the systemsperfor- ic signalat a distantfarfield point in an obliquelybackward direction is a mance.Such checks have traditionallybeen limited to the laboratorybe- sum of N terms,each of the genericform A, cos(coot)d- B, sin(coot), causethey requirespecial acoustic environments. A new devicefor whereA, and B, dependon the instantaneousposition of the nth scat- calibratingand checking intensity systems will bcconsidered here. This terer. Sincethese positions change with time, the coe•cientsalso vary, deviceconsists of an acousticsource and a coupler.The couplercan be althoughslowly over intervals comparable to l/coo.Statistical properties configuredin two ways.In oneconfiguration both microphones arc ex- of the timevarying sum functions A (t) andB(t) arestudied analytically posedto the samesound field. This mode can be usedto calibratethe and with numerical simulations for various models of the sound wave and channelsensitivities and to measurethe residualintensity index of the theBrownian motion. A principalproblem adressed is that of determining system.In theother configuration the coupler will simulatefree-field con- the statistics of the time interval At over which a fixed number K of zero ditions.Either a pistonphoneor a broadbandsource can be usedas the crossingsoccur for the sum A(t) cos(coot)d-B(t) sin(coot),given that acousticsource. The designof the devicewill be consideredand its use this intervalis centeredat a particulartime, the pertinentquantity of illustrated. interestbeing the relative error in theaccumulative phase difference 2•rK / At - cooper unit time.The resultssupport the generalconclusions that smallererrors are expectedfor higheramplitude sound waves and for soundwaves of lowerfrequency. [Work supportedby ONR.] 11:00
EEl0. A multimicrophonesystem with spatialselectivity. Hikaru Date (Departmentof InformationEngineering, Yamagata University, 4-3-16, 11:30 Johnan, Yonezawa, 992 Japan) and Ken'ichi Furuya (Acoustic Department,Kyushu Institute of Design,4-9-1, Shiobaru,Minami, EEl2. Differences in measured microphoneresponses at different Fukuoka,815 Japan) distances from artificial mouth. Frederick M. Kruger (Kruger The principleof spatialselectivity of a receivingsystem, which com- Associates,37 SomersetDrive, Cornmack,NY 11725), Mitchell Mayer (Sonetronics, Inc., 1718 H Street, West Belmar, NJ 07719), prisesseveral microphones distributed on a sphericalboundary surface John Bareham (Bruel & Kjaer, Marlboro, MA 01752), and andone microphone at thecenter of thespace enclosed by theboundary, is introducedin accordancewith thetheorem expressed by Kirchhoff'sinte- Barbara Kruger(Kruger Associates, 37 SomersetDrive, Cornmack,NY 11725 ) gral equation.The output from eachof the boundarymicrophones are processedin thefollowing way. Each signal is firstdelayed and then differ- Whencalibrated and operatedat differentdistances, available artifi- entiatcdin both time and spacein the directionnormal to the spherical cialvoice/mouth frequency response measuring systems yield significant- boundary.All of thedelayed and differentiated signals are added together ly differentresults. These differences are greaterfor differentdistances with appropriateweighting. The systemoutput is finallyobtained by sub- than for differentvoice/mouths. Virtually all reportedvoice measure- tractingthe weighted sum from the output of thecenter microphone. The ments are made at the CCITT recommended distance of 25 mm from the systemoutput contains no componentsof externalnoise from the outside lips,or 25 mmin frontof the "lip ring"of theartificial voice/mouths. Yet, of thespherical boundary. Simulation studies were carried out for various all high-noiseenvironment military microphonesare tested i/4 in. (6 numbersof microphonesand for differentratios of signalwavelength to mm) from the voice/mouthopening. This is, however,several millime- the radiusof the sphericalboundary and for both plane and spherical tersbehind the lip planeof the voices.At this distance,noise canceling incident waves. microphonesshow minimal cancellation effects. Historically, the useof 1/4-in. distancebegan with a laboratoryartificial voice/driver sound sys- tem, which predatedthe first commerciallyavailable, relatively large 11:15 scaleproduction artificial voice (B&K 4216). This papershows the feet(s) on measuredmicrophone response when military noise canceling EEl1. Absolutecalibration of acousticsensors utilizing electromagnetic microphones(M-138, H-250,M-162) anda calibratedlaboratory micro- scatteringfrom in situ particulate matter. Allan D. Pierce and Yves phoneare tested,on axis,at differentdistances from severalcalibrated H. Berthelot(School of MechanicalEngineering, Georgia Institute of voices(6 mmfrom opening, 6 and25 mm in frontof lip ring). Realworld Technology,Atlanta, GA 30332) implicationsare discussed.
$64 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S64 THURSDAY MORNING, 19 MAY 1988 METROPOLITAN BALLROOM, 8:00 A.M. TO 12:00 NOON
SessionFF. SpeechCommunication VI: Topicsin SpeechPerception (Poster Session)
Fred D. Mini fie, Chairman Departmentof Speechand HearingSciences, University of Washington,JG-15, 1417N.E. 42ndStreet, Seattle, Washington98195
ContributedPapers
All pasterswill bedisplayed from 8:00 a.m. to 12:00noon. To allowcontributors the opportunity to seeother posters,contributors of odd-numberedpapers will beat their postersfrom 8:00to 10.00a.m. and contributors of even-numberedpapers will be at their postersfrom 10:00a.m. to 12.00noon.
FFI. Consonantand vowel intelligibility with the Symbioneoehlear metricfunctions revealed that infantsrequired a morefavorable S/N than implant.Michael F. Dotman (Departmentof Speechand Hearing adultsto achievegiven levels of performance.The adult subjects and 14of Science,Arizona State University, Tempe, AZ 85287) the 16infants were also tested using a discrimination-thresholdprocedure in which the stimulusintensity was variedadaptively (l-up/l-down). Studiesof consonantintelligibility ( 16items in aCaformat) and vow- The adaptivediscrimination procedure provided, for both groups,esti- el intelligibility( 12items in bVt format)were conducted with sixpatients matesof performancethat are in goodagreement with thosefrom the fittedwith theSymbion four-channel eoehlear implant. Consonant intelli- perfotnnanceversus SfN functionsthat weredeveloped using the more gibility rangedfrom 52% to 81% correct.Voicing and mannererrors time-consumingmethod of trial blocksat fixedS/N's. [Work supported wereuncommon. Vowel intelligibilityranged from 22% to 79% correct. by DRF and NIH POI NS!6337.] Errors were usuallyvowels with similar F 1 and/or F2. For the bestpa- tient, an analysisof errorsas a functionof channelstimulated indicated that 81% consonantintelligibility could be achievedwith the useof two functional channels,one centeredat 500 Hz and the other centeredat 4000 I-tz. FF4. Relationof children'sidentification of consonant-vowelsyllables to their fine-grainedauditory discrimination:A progressreport. Lois L. Elliott, Michael A. Hammer, Margo E. Scholl, and Bonnie S. Anthony(Northwestern University, 2299 Sheridan Road, Evanston, IL 60208) FF2. Speechperception: The effectsof recurrentconductive hearing loss in youngchildren. Richard L. Clarkson (Rhode Island Schoolfor the Just-noticeabledifferences (JNDs) and identification(i.e., labeling} Deaf, Corliss Park, Providence, RI 02908), Peter D. Eimas functionswere obtained for childrenaged 6 through11 years. Two conso- (Departmentof Psychology,Brown University, Providence, RI 02906), nant-vowel(CV} continuarepresenting the place-of-articulationand andG. CameronMarean (Department of Speechand Hearing Sciences, voice-onset-timefeatures served as stimuli;the adaptivediscrimination JG-I 5, Universityof Washington,Seattle, WA 98195) taskthat has been previously described [e.g., L. L. Elliottet aL, J. Acoust. SOe.Am. 70, 669 ( 1981) ] wasused. Some children produced consistent The ability of five-year-oldchildren with historiesof recurrentotitis identificationfunctions for both continua;some were inconsistentin la- media(OM) andassociated conductive hearing loss to perceivesyllable belingCVs of onecontinuum only; and some were inconsistent for both initialstop consonants was investigated. Naturally produced, computer- continua.In manycases, the childrenwith the poorestidentification per- editedstimuli were created to formcontinua that rangedin voiceonset formancealso exhibited the poorestJNDs for the samestimulus contin- time(VOT) from7 to 60ms. The ability of children to assign these stimuli uum.The irregularidentification functions observed for someof these to adultphonetic categories was assessed by identificationand discrimina- normal-hearingchildren resembled functions that have been reported for tion tasks.Three subject groups were identified: One group (group C) someadults [B. E. Walden et aL, J. Acoust.SOC. Am. 79, 1101 (1986)]. had no historyof OM and waslinguistically age appropriate; another [Work supported,in part, by NINCDS (NIH).] group (groupOM) had considerableinvolvement with OM and wasalso linguisticallynormal; the third group(group OMDL) hadan equalin- volvementas group OM withotitis media, but was linguistically delayed. All groupswere matched for age, IQ, socioeconomicclass, and education- al experience.Results indicate that OM wasassociated with poorer dis- FFS. Normativedata and physicalmeasurements for a nonsenseCVC criminationscores, regardless of thelinguistic status of theindividual. intellig•bilitytest. ChristineM. Rankovicand Dianne J. Van Tasell (Departmentof CommunicationDisorders, University of Minnesota, Minneapolis,MN 55455) A nonsenseCVC stimulusset has been developed for investigations FF3. Infant speech-•ounddiscrimination in noise. Robert J. Nozza, intothe application of thearticulation index (AI) to speechrecognition LindaC. Bond,Sandy L. Miller, and RevaN. F. Rossman(Division of testingwith hearing-impairedpersons. Each test list consists of 57 items Audiology,Children's Hospital of Pittsburgh,One Children'sPlace, drawnat randomfrom a corpusof 228items comprising syllables spoken Pittsburgh,PA 15213) byfour talkers. Crossover functions support the similarity of thisstimulus setto that of Frenchand Steinberg [ J. Acoust.Soc. Am. 19, 90 (1947) ] in Infantability to discriminate a speech-sound pair in noisewas assessed termsof frequencyimportance function. Because the number of vowels using the visual reinforcement infant speechdiscrimination procedure. wasrestricted to three,the long-termspectrum of the stimulusset is less Sixteeninfants between7 and 12 months were teatedon the foa-ga/ smooththan that usuallyassociated with connectedspeech. The AI ver- contrastat four S/N's ( -- 8, 0, 8, and 16 dB). A groupof young-adult suspercent-correct phoneme score functions have been obtained from controlsubjects was tested also at four S/N's ( -- 12, -- 8, -- 4, and 0 normallyhearing subjects. Comparisons with publisheddata for similar dB). The noisewas continuously present at 48 dB SPL. Group psycho- stimulussets will be made.Talker differences,reliability of subjectperfor-
S65 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115thMeeting: Acoustical Sociely of America S65 mance,and preliminary data from hearing-impaired subjects will bere- formsobtained from the diplophonicsubjects. In the frequencydomain, ported.[Work supportedby NICHHD T32 HD-07151and NINCDS NS the diplophonicvoice spectrum bears little resemblanceto that of normal 12125.] phonation.Instead of the normalset of harmonicpeaks at integermulti- piesof thefundamental, the diplophonic spectrum shows a wealthof very closelyspaced energy peaks of whichthe two fundamentalsare often the mostprominent but rarelythe lowestfrequency. The spectralenvelope of FF6. Patterns of phonemeconfusions with senseand nonsenseCVC theglottal area waveform may appear to havepoles where no formants are words for subjects with normal hearing and with presbycusis. possibleand the spectralenergy falls off quite rapidly. The locationsof Arjan Bosman and Guido F. Smoorenburg (Laboratory of prominentenergy peaks were predicted by modelingthe displacementof Experimental Audiology, Department of Otorhinolaryngoiogy, the vocalfolds from midlineusing the sumof two sinewaves clipped by UniversityHospital, Catharijnesingel101, Utrecht 3511 GV, The pulsetrains to simulatecollision effects of the two vocal folds. [Work Netherlands) supportedby NINCDS. ] Differencesin listeningstrategy between normal-hearing and hearing- impairedlisteners will becomeapparent in differentpatterns of phoneme confusions.The confusionswere studied with meaningfuland meaning- lessword material of theconsonant-vowel-consonant type (CVC words) FF9. Elicitation of spontaneousspeech production from preschool-aged presentedin quietat fixedpresentation levels to a groupof youngnormal- tactile aid users.Adele Proctor (NortheasternUniversity, Boston, MA hearingsubjects, a groupof elderlysubjects with normalheating regard- 02115) and Corine Bickley (MassachusettsInstitute of Technology, ing their age,and subjects with presbycusis.The confusionsof the vowels Cambridge,MA 02139) andinitial andfinal consonants were subjected to INDSCAL analysis[J. Engineeringof sensorysupplements that are sufficientlysmall in size, D. Carroll andJ. J. Chang,Psychometrika 35, 283-319 (1970) ]. Param- lightin weight,and adequate protocols for measuringdevice efficacy have etersin the analysiswere; word type (meaningfulversus meaningless), longbeen problems for thoseinvolved in designingvibrotactile aids for groupof subjects,and phoneme identification score. This latterparameter preschool-ageddeaf children.Contributing to the engineeringvariables is perceptuallymore relevantthan presentationlevel whencomparing influencingtactile aid development are the problems of howto elicitrepre- subjectswith differenthearing losses. For the normal-hearingsubjects, sentativesamples of spontaneousspeech production before, during, and vowelperception was dominated by the first and second formant, whereas after tactileaid usageand how to quantifyeffects of tactileaid usageon for presbycusissubjects the contributionof the secondformant was re- speechproduction of childrenwho exhibitnonnormal articulatory pat- ducedand some influence of vowelduration was present. For the percep- terns.Since visual displays presented on microprocessorsare reportedto tion of bothinitial and final consonants voicing was all important.How- be effectiveelicitors of vocaloutput, available databases on articulatory ever,the featureof voicingwas better perceived by presbycusissubjects and acousticcharacteristics of hearingand hearing-impairedpres- thanby the normal-hearingsubjects. choolers,primary acoustic parameters known to influenceintelligibility, and availabledata suggestingeffects of tactileaids on speechproduction wereincorporated to designgraphic stimuli to elicitspeech from 3- to 5- year-olddeaf tactileaid users.The purposeof this postersession is to FF7. Relations betweenthe intelliglbility of sentencesand words for discussthe theoretical basis for selectionof stimuli, to demonstrate the subjects with normal hearing and with presbycusis. Guido protocolfor eliciting spontaneous productions, and to presentthe planned F. Smoorenburgand Arian Bosman (Laboratory of Experimental articulatoryand acousticmeasurements. [Work supportedby DOg., Audiology,Department of Otorhinolaryngology,University Hospital, Grant No. 133 GH70189.] Catharijnesingel101, Utrecht 3511 GV, The Netherlands)
In order to seewhether hearing-impaired listeners make more exten- siveuse of nonacousticcues than listenerswith normalhearing, different speechmaterials were presented to threegroups of subjects.The speech FF10. Acoustic characteristicsof dysarthria associatedwith cerebral materialconsisted of meaningfuland meaninglesswords of the conso- palsy. Gary Weismet and Karen Forrest (Speech Motor Control nant-vowel-consonanttype (CVC words)and short, simple sentences of Laboratories,University of Wisconsin--Madison,Madison, WI 53705) eightto ninesyllables [R. P!ompand A.M. Mimpen, Audiology18, 43- 52 (1979)]. All material was uttered by a male and a female speaker. This paperwill reporton qualitativeand quantitative observations on Three groupsof subjectsparticipated in the experiments:young normal- the speechproduction characteristics of adultswith cerebralpalsy. The hearingsubjects, elderly subjects with goodhearing regarding their age, subjectswere ten adultswith cerebralpalsy, primarily of the spastictype. and subjectswith presbycusis.Test-retest reliability for both word and Multiple repetitionsof selectedsentences were analyzedacoustically to sentencematerial was found to beclose to the theoreticalvalue. Averaged determinethe spectraland temporal characteristics of segments.In addi- acrossall groupsof subjects,the correlationbetween the SRT for word tion, certainqualitative analyses, involving the form of spectrographic andthe SRT for sentencematerial was 0.97, whichresulted in a prediction displaysacross utterance repetitions, were performed to addressthe issue error of 5.3 dB. Predictionof the SRT's from the toneaudiogram with the of randomversus consistent control of articulatorygestures in nenrogcnic Fletcherindex (average heating loss at 500, 1000,and 2000 Hz) resulted speechdisorders. It is arguedthat in somecases a patientmay produce in a predictionerror of 5.5 and6.8 dB for the SRT for wordand sentence very aberranttemporal and spectralfeatures which, nevertheless,are material,respectively. quiteconsistent across repetitions.
FI•. gource cl•recterlst•cs of dlplophonle. Bruce R. Gertart, FFII. Aeoustle characteristics of speechproduced by an older geriatric Kristin Precoda, David G. Hanson, and Gerald S. Berke (Veterans population. Julie M. Liss, Gary Weismet (Department of Administration Medical Center, West Los Angeles, Wilshire and CommunicationDisorders, 1975 Willow Drive, Universityof Wisconsin, SawrelicBoulevards, W 126,Los Angeles, CA 90073and UCLA Schoolof Madison, WI 53706), and John C. Rosenbek (Middleton Memorial Medicine,Los Angeles,CA 90024) VeteransAdministration Hospital, Madison, WI 53705) Diplophoniadescribes the perceptionof more than one fundamental The effectof advancingage on speechproduction characteristics is not frequency componentin a voice. Simultaneousphotoglottographic clearly understood.Yet it is crucial for those who study and serve the (PGG), electroglottographie(EGG), and acousticsignals were collect- normaland neurologicallyimpaired geriatric populations. The purposeof ed from five diplophonicsubjects. Results indicate that dipiophoniais thisstudy was to obtaina normafivedatabase for the speechproduction similarto the beatfrequency phenomenon caused by independentvibra- characteristicsof an oldergeriatric group. Fourteen veterans, 87 to 93 tion of the two vocalfolds at slightlydifferent frequencies. In the time yearsold, servedas subjects.A total of 40 sentenceswere produce at a domain,modeling vocal fold displacementfrom midlineby summingtwo conversationalrate by each gentleman.Wideband (300-Hz) spectro- sinewaves produced area waveforms closely resembling the PGG wave- gramswere created from high-quality tape recordings. Specified acoustic
S66 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S66 measureswere made (consonant,vowel, and voiceonset time durations, hearing95 dB SPL pink noise.The sentenceswere digitized, segmented, and vowelformant trajectories) via a hit-paddigitizer and microcom- and transcribedusing SPreE. The resultingdatabase consisted of approxi- puterSOhogram analyzer program. Descriptive analysis was performed to mately850 segments per subject per speaking condition. The distributions obtainnormal parameters of speechproduction. Current theories of aging of spectraland temporalproperties of classesof segmentswere deter- andneurological diseases affecting speech will bediscussed. minedusing SEARCU. All subjectsproduced sentences in the presenceof noisewith increasesin fundamentalfrequency and total energy,as has beenfound for isolatedwords. Segment durations and spectral character- isticswere affectedby noisefor somesubjects more than for others. FFI2. Analysisof formanttransitions for Englishfricatives spoken by [Work supportedin partby AFOSR.] five male and five female talkers. Kelsey Key and Dennis H. Klatt (Room 36-523,Massachusetts Institute of Technology,Cambridge, MA 02139)
Nonsense words of the form C'VC'¾C'VC were recorded for the four FFI$. Fundamentalfrequency contours of prewordvocalizations through voicelessfricatives of English[f, th,s,sh]in ten vowel environmentsby the emergenceof wordcombinations. Michael P. Robb (Departmentof eachof five male and five femaletalkers. The databasewas digitized at SpeechPathology and Audiology,University of Hawaii, Honolulu,HI 16 000 samples/sand digitalspectrograms were produced. Formant fre- 96822) and John H. Saxman (Communication Sciencesand Disorders quencyvalues, F2 andF 3, weremeasured by handfrom the spectrograms Program,Syracuse University, Syracuse, NY 13244) at voicingonset and vowel midpoint for eachCV transition.These data An increasingnumber of acousticcharacteristics of infantpreword wereexamined in orderto seekregularities of severalpossible forms: ( 1) vocalizationshave been found to besimilar in manyrespects to the vocal- invariantonset frequencies independent of speakerand vowelenviron- izationcharacteristics of children'searly meaningfulspeech. Such find- ment,(2) existenceof locusfrequencies that differas a functionof placeof ingsare suggestiveof continuityin the underlyingprocesses involved in fricativearticulation, but are similar acrossspeakers, and (3} similar acquiringlanguage. To date,few studies exist that have examined in detail transitionbehavior (e.g., rising or fallingF2 and/or F3) acrossspeakers the acousticcharacteristics of babbling,early words, and word combina- for a givenvowel with no attemptto seekregularities across vowels. In all tionsin a singlegroup of children.The presentstudy providesmonthly cases,distributions of transition-relatedmeasures across speakers for data on fundamentalfrequency (F0} contoursin seveninfants who were eachplace of articulationshow disturbing amounts of overlap,even when audiotape recorded over a 12-monthperiod. Within a year'stime, each maleand femaledata are plottedseparately. While it hasoften been said child'scommunication development progressed from babbling vocaliza- that [f] and [th ] aredistinguished perceptually by F2 andF3 transition tionsto wordcombinations. The frequencyand typeof F0 contoursin behavior(because frication spectra are oftenquite similar), thedata sug- eachchild's mono-, bi-, and polysyllabicvocalizations are reported.Re- gestthat it will be difficultfor a speechrecognition device to makethis gardlessof syllabletype, the children used a largenumber of rising-falling distinctionreliably from observed formant transition behavior in thegen- F0 contoursacross preword and multiword periods of vocalizationdevel- eral population.Some reassessment of the appropriateperceptual cues opment.The leastfrequently occurring F0 contoursin bothgroup and seemsto be required.[Work supportedin partby an NIH grant.] individualdata werefalling-rising contours. It is proposedthat unifor- mity existsin early vocalizationsacross preword and meaningfulspeech periodsdue to physiologicallymotivated factors.
FFI3. Aeonstieproperties of isolatedstop consonant release transients• BrunoH. 'Repp (HaskinsLaboratories, 270 Crown Street,New Haven, CT 06511-6695)and Hwci-Bing Lin (HaskinsLaboratories, 270 Crown FF16. Classification of vowels using spectrum moments. Street, New Haven, CT 06511-6695and Departmentof Linguistics, Paul Milenkovic and Karen Forrest (Waisman Center, University of Universityof Connecticut,Storrs, CT 06268) Wisconsin,1500 Highland Avenue, Madison, WI 53705-2280) This study focuseson the initial componentof the stop consonant Formantfrequencies form a setof acousticfeatures that can be used to releaseburst, the releasetransient. In theory,the transient,because of its classifyvowels phonetically [A. K. Syrdaland H. S. Oopal,I. Acoust. impulselikesource, should contain moch informationabout the vocal SOc.Am. 79, 1086-1100(1986) ]. The advantageof formantfrequencies tractconfiguration at release,but it is usuallyweak in intensityand diffi- isthat theyare unaffected by changesin the speechspectrum slope attrib- cult to isolatefrom the accompanying frication in naturalspeech. For this utableto the voicesource. The disadvantageof formantfrequencies is in investigation,a human talker producedisolated release transients of thelarge errors in frequencythat can result from a misclassificationof the fo,d,g/in nine vocaliccontexts by "mouthing"these syllables. He also formantpeaks of the speechspectrum. A setof featuresbased on statisti- producedthe corresponding CV syllableswith regularphonation for enm- cal momentscomputed from the speechspectrum has been successful in parison.Spectral analyses showed the isolated transients to havea clearly classifyinga broad class of obstruentsounds [K. Forrestet aL, J. Acoust. defined formant structure, which was not seen in natural releasebursts Soc.Am. Suppl.i 82, S84 (1987) ]. In the caseof vowelsounds, moment whosespectrum was dominated by the frieationnoise. The formantfre- analysiseliminates the problemof identifyingformant peaks but intro- quenciesvaried systematically with both consonantplace of articulation ducesthe problemof variabilityin the momentfeatures resulting from and vocalieenntext, with the latter, "coartieulatory"influence actually changesin the voicesource. In orderto correctfor betweensubject vari- exceedingthe former. The transientthus provides an acoustic snapshot of ation in the voicesource, the long-termaverage spectrum of all of the the vocal tract immediatelyafter the release. [Work supportedby speechsamples recorded from each subject is computed and used to nor- NICHD.] realizevowel spectra prior to computingmoments. Between subject clas- siftcationresults for adult subjectsare presented.[Work supportedby NIH grantsNS 21516and NS 13274.]
FFI4. Somephonetic characteristics of sentencesproduced in noise.Z. S. Bond (Departmentof Linguistics,Ohio University,Athens, OH 45701), Thomas]. Moore (ArmstrongAerospace Medical Research FFI7. Factor annlysis for vowel and frlcative duration in American Laboratory,Wright-Patterson AFB, OH 45433), and Kate McCreight English.John F. Pitrelliand Victor W. Zue ( Room36-575, Department (Departmentof Linguistics,Massachusetts Institute of Technology, of ElectricalEngineering and Computer Science, Massachusetts Institute Cambridge,MA 02139) of Technology,Cambridge, MA 02139) In previouswork, relativelysystematic differences were found in the In an attemptto developa comprehensivemodel of segmentduration phoneticstructure of spondeewords produced in isolationand in the for AmericanEnglish, the durationof vowelsand fricativesas a function presenceof noise.The purposeof this studywas to determinewhether ofsegmental and suprasegmental factors was studied. Two databaseswere noiseaffects continuous speech in a similarmanner. Four speakerswere used,one consisting of onereading of the 720 Harvard List sentencesby recordedreading 20 sentences,two timesin quiet and two timeswhile one male speaker,and anotherconsisting of the first 500 Harvard List
S67 J. Acoust.Soc. Am. Suppl. 1, VoL 83, Spdng 1988 115th Meeting:Acoustical Society of Amedca S67 sentences,with eachlist of ten readby onemale and onefemale speaker. manceof dichoticlistening and brain-damaged subjects indicates that the Segmentdurations were obtained from time-alignedphonemic transcrip- righthemisphere is specializedfor recognizingfamiliar voices (the right tions.Factors influencing duration and theirinteraction effects were stud- hemisphereis believedto be superiorin patternrecognition); (3) in ied usinganalysis-of-variance techniques. Results indicate that voweldu- pairedcomparison tasks, normal subjects compare discrete auditory fea- rationsare largelyaffected by factorssuch as vowelclass (diphthongs, tures,as well as the overall pattern; (4) brain-damagedsubjects who show tense,lax, and reducedvowels), sentence-final lengthening, the syntactic impairedperformance in unfamiliarvoice discrimination have damage to functionof the word in which they appear,and lexicalstress. Other fac- eitherthe right or left temporallobe (specialized for processingof audi- torspreviously thought to beimportant, such as the voicing characteristic tory detail). Testablehypotheses generated by this modelwill alsobe of thefollowing consonant, did notaccount for muchof thevariance in the discussed. data.Using five of the mostsignificant factors, an accountwas obtained for 67% of the varianceon trainingdata as well as on test data for the single-speakerdatabase, and 63% and 59% for the multispeakerdata- base.The differencein resultsfor the two databasesmay be attributableto FF21. Speechrate and genderas mediatingfactors in personperception. interspeakerdifferences that includespeaking-rate variations. In an oral Faith-Anne Dohm, Stanley Feldstcin, and Cynthia L. Crown presentation,the resultson the modelingof fricativeduration will bedis- (Departmentof Psychology,University of Maryland,Baltimore County, cussed.[Work supportedby DARPA under contract N00014-82-K- 5401 Wilkens Avenue, Baltimore, MD 21228) 0727, monitoredthrough the Officeof Naval Research.] A previousstudy by the authors found that judgment of speechrate is independentlyinfluenced by the gender of thejudges and of thespeakers. The researchof others[e.g., B. Brownetal., J. Acoust.Soc. Am. 54, 29- 35 (1973)] hasshown that speech-rate perception mediates other aspects FFIS. Spectrographiccomparison of whisperedvoiced and voiceless stop of interpersonalperception. The study reported here, then, investigated consonantsin variousvowel environments.Igor V. N•b•lek, Hsiao- the possibilitiesthat: (a) Individualswho speak rapidly are viewed in Chuan Chen (Department of Audiology and SpeechPathology, morepositive ways than those who speak slowly; (b) malesarc viewed in University of Tennessee, Knoxville, TN 36996-0740), and morepositive ways than females; and/or (c) speechand gender jointly SumalaiMaroonroge (Speech and Hearing Clinic, Ramathibodi influenceperson perception. Seventeen male and 28 femalecollege stu- Hospital,Bangkok 10100, Thailand) dentslistened to and rated, eachof ten adjectivescales, three male and A previouspresentation reported on spectra of whisperedand phonat- threefemale speech samples. An hierarchicalmultiple regression analysis ed speechand on comparisonof identificationof whisperedvoiced and indicatedthat, whereasthe malespeakers received less positive ratings voicelessconsonants when those consonants were presented in C/a/sylla- whenthey spoke rapidly, the female speakers received less positive ratings bles.The cuesfor voicingin whisperedplosives were studied using spec- bothwhen they spoke rapidly and slowly. The results suggest that person trograms.In a follow-upstudy, spectrographic analysis of whispered perceptionisinfluenced jointly by the speech rate and gender of a speaker. monosyllableswith plosivesin variousvowel environmentswas per- formed.The resultsof the analysis--withrespect to the voicingdiffer- ences---willbe presentedand discussed. FF22. Turning the teaching of foreign pronunciationsinto a computer product.Raymond J. Thomas(Facult• desSciences de Luminy, 13288 Marseille Cedex 9, France) FF19. Voice"features" in long*termmemory. Jody Kreiman (Phonetics Laboratory,Department of Linguistics,UCLA, LosAngeles, CA 90024) Extendingthe scientificteaching of Englishand Frenchpronunci- andGeorge Papeun (Los AlamosNational Laboratory, Computer User ations[J. Acoust.Soc. Am. Suppl.I 76, S90 (1984) ] to adultsspeaking ServicesC-10, Mail StopB-296, Los Alamos,NM 87545) otherlanguages showed that specificvariations in speechacoustics and useof musclesare relatedbasically in the samemanner in all humans,but This studyexamines the perceptualparameters used to recognizean that an addedcomplexity arises from seemingly unrelated speech habits unfamiliarvoice (heard only once before) after a one-weekdelay, as com- dependingon eachmother tongue. Final results of adviceare thus unpre- paredto theparameters used to discriminateamong the same voices in a dictable,but practicehas shown that progresson variouspoints could be short-termmemory (paired comparison) task. Ten groups of tenlisteners totaled throughgradual classifications of oppositions.This could bc eachheard a shortsample of a singlevoice; after seven days, all listeners achievedwith a computerprogram using the speechsyllabic data bank heardthe full ten voiceset and had to determineif theirtarget voice was (Proc. Xlth ICPhS, Tallinn, 1987, Sec. 89.4) togetherwith a much presentor not.Confusion data and similarity ratings were gathered and smallerbank on rhythmsand stresses in the variouslanguages. The study combinedto form full matricesthat were analyzedusing multidimen- of fewerthan 20 phoneticsystems would cover all importantspeech hab- sionalscaling. Differences in theperceptual parameters that underlie con- its.This wouldresult in thecapacity for thecomputers to "think out," in a fusionsand perceived similarities in theshort- versus the long-termmem- matter of minutes,the one amongthe 16 000 000 odd teachingprograms ory taskswill be discussed,as will ½hanõesin the relativeimportance of it could devise(Evaluation dc I'information cn langucsnaturelies, Ms dc differentparameters with time. The implicationsof thesefindings for Th•se, BU de Rouen,1983, Vol. 3 ) that wouldbe best adapted to any.one modelsof voiceperception and speaker recognition will alsobe addressed. of usfor learningany of the 4000 odd foreignpronunciations.
FF20. A neuropsychologicalmodel of voice perception. Diana FF23. Time-gatedword identification performance. Chic Higuchi Craig Van Lancker (PsychiatryDepartment, UCLA, Los Angeles,CA 90024) (Department of SpeechPathology and Audiology, University of and Jody Kreiman (PhoneticsLaboratory, Linguistics Department, Wisconsin-Milwaukee, Milwaukee, WI 53201 ) UCLA, Los Angeles,CA 90024) Listenerswho are able to identifywords from their earliestonset dem- It will beargued that two neuropsychologicalprocesses mediate voice onstratean effectiveintegration of acoustic-phoneticand predictive-con- perception:pattern recognition and featural analysis. Properties of voices textualcues for real-timespeech recognition. In this investigation,a gat- that determinethe relativedominance of thesetwo kindsof processing ing paradigmwas used to studyhow presentationtime (durationfrom include:( 1) the familiarityof thevoices, and (2) the hetcrogeneityof the word onset) influencesword identificationperformance. A digital sona- voice set--how "similar" the voices arc to each other. Familiar voices that graphwas programmed to storeand time gate50 isolatedmonosyllabic differ substantiallyare processedlargely as complexpatterns, while a wordsdrawn from a recording(Auditee) of the NU-6 (list IA). Twenty homogeneoussetof unfamiliarvoices recruits some featural analysis. This normal-hearingyoung adults were testcA monaurally via carphones. Each modelis derivedfrom the followingobservations: ( 1 ) Acousticallyalter- wordwas presented repeatedly with presentationtimes increasing in 60- ing familiarvoices affects their recognizabilityunevenly, suggesting that ms incrementswith each successivepass. Words were presentedin 3-7 no fixedset of parametersunderlies the recognitionprocess; (2) perfor- passesuntil eachword was completely presented ( 180-420ms). Subjects
$68 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115thMeeting: Acoustical Society of America S68 wereasked to guessthe word,write it down,and indicate how confident benchmarkofdifferentiability of intact sentence pairs, the perceptual dis- theywere about each guess. Results reveal the effectof a 60-msgating tinctivenessofspontaneous and read utterances was evaluated by deleting paradigmon word identification performance and relations between word thesegmental information alone by low-passfiltering. In additionalcon- duration,listener confidence, and percentof wordsidentified. vergingtests, versions of the sentencepairs without formant variation weresynthesized, setting the spectrum to thevowel [ o], whilepreserving naturaldifferences in durationand f0; and,synthetic sentence pairs lack- ing variationinfo weretested in filteredand [ o ] versions.The resultswill FF24. Spectral shaping and the intelliglbility of speechin noise. bediscussed in referenceto theperceptual constraints on thecomprehen- R. Pedlow (SRL/Psychology Department, Indiana University, sion of spontaneousspeech. [Research supported by grants from Bloomington,IN 47405) NINCDS and NICHHD. ]
Various studies have shown that attenuation of the first formant re- sultsin enhancedspeech intelligibility in noise.One explanation(Nye et al., 1974) suggeststhat it is causedby upwardspread of maskingin nor- FF27. Learning effects for two different speech synthesizerswith mal speechperception. The otherexplanation proposed is that attenuat- differentspeech material. Rory A. DePaolis(The GraduateProgram in ing F 1 increasesthe amountof intelligibilityinformation relative to the Acoustics,The PennsylvaniaState University, University Park, PA total speechenergy in the signal.The stimulusmaterials were phonetical- 16802 ) ly balanced(PB) wordsproduced by one male speaker.Stimuli were digitallyfiltered with a high-passfilter that had a passbandfrom 1100- This studyexamined the relationshipbetween the learningeffects due 5000 Hz and a stopbandfrom 0-850 Hz (the attenuationlevels of the to two inexpensivemethods of synthesizingspeech, formant synthesis stopbandwere 10, 20, 40, and80 dB relativeto theunfiltered signal. Items (FS) and linear predictivecoding (LPC). An experimentusing two wereequated for rms energyafter filtering.Processed stimuli were pre- groupsof sixsubjects listening to eitherFS or LPC for four consecutive sentedto subjectsin a perceptualidentification task at 0 dB S/N. Articu- dayswas conducted. The test materialconsisted of contextualand non- lationindex values were then computed for the testitems. Fitting transfer sensesentences with written clues,contextual sentences without written functionsto the datashowed that the increasedintelligibility obtained by clues,and the rhymetest. A prosepassage was also presented to the sub- high-passfiltering was not well predictedby the articulationindex values. jects.There was no significantdifference between the learningcurves for Thissupports the hypothesis that attenuating the F 1 resultsin an increase the two synthesizersbut there was an overalldramatic learning effect, in intelligibilitydue to a releasefrom maskingeffect. mostnoticeably from the first to third presentationacross all typesof speechmaterial. The synthesizersexhibited equality of meanpercent cor- rect word scoresfor all materialover the four sessions,excluding day 1, usingthe nonsensesentences in whichthe FS devicewas more understan- FF25. Perceptual evidence for stress foot structure. Jan Edwards dable.These results are presentedgraphically with comparisonsto other (Hunter College,425 East 25 Street, New York, NY 10010) and studiesusing natural and synthetic speech and predictions on theperfor- Mary Beckman(Department of Linguistics,The OhioState University, manceof the two synthesizersin noise. Columbus, OH 43210) A yearago, Beckman and Edwards [J. Acoust.Soc. Am. Suppl.1 81, S67(1987) ] reportedon twodurational effects. Phrase-final lengthening FF28. Perceptionof digJtullycoded speech under some cognitive load. isa phonologicmark ofintonational phrase boundaries: It isa substantial Kazunori Ozawa (C&C Information Technology Research effectwhose domain is highlyconsistent across speakers. Seeming word- Laboratories, NEC Corporation, 4-1-1, Miyazaki, Miyamae-ku, final lengtheningis a much smallerand more variableeffect--how to Kawasaki213,Japan) and JohnS. Logan(Speech Research Laboratory, describethe domain of this smaller effect, or even whether it should be Departmentof Psychology,Indiana University, Bloomington, IN 47405) describedphonologically, was unknown. The perceptual salience of these two effectshas beenexamined, reasoning that if they are phonologic Perceptionof meaningfulsentences that were encoded by speechcod- marksof prosodicconstituent boundaries, then listeners should be able to ing methodswas measured. In orderto differentiateand to expandper- perceivethe edgesso marked. Listeners can consistently use intonational ceptualdifferences between the speech coding methods for the meaningful phrase-finallengthening, but not the smallerdurational effect, to dis- sentences,a task of recall for the sentencesunder cognitiveload was criminatebetween phrases that are otherwise identical. It issuggested that adopted.In thisexperiment, the distractor method of short-termmemory thesmaller effect is nota phonologicmark ofa prosodicconstituent with- [L. R. Petersonand M. J. Peterson,J. Exp.Psychol. 58, 193-198(1959) ] in the intonationalphrase. Rather, it is a resultof articulatorytiming wasutilized as the cognitiveload to subjects.Pitch predictive multipulse relationshipswithin the stress foot, with the longer[pa] in 'popopposed' speechcoding at 8 kb/s [ K. Ozawaand T. Araseki,Proc. ICASSP, 1689- ascompared to 'poppaposed' due to their differentstress feet structures. 1692(1986) ] and50 kb/s/•-lawPCM wereused as speech coding meth- [Work supportedby NSF Grants.No. IRI-8617873 and IRI-8617852.] ods.As a reference,uncoded natural speech was also used. A total of 60 Harvard sentenceswas usedas meaningfulsentence stimuli. Results showedthat the higher the cognitive load, the larger the perceptual differ- encesbetween uncoded and codedspeech, as well as betweenMPC and PCM. However, differencesin error rates among uncodedand coded FF26. Do segmentalattributes distinguishspontaneous and prepared speechwere very small in thecase of no cognitive load. The results suggest speech? Robert E. Remez, Anne Sherman, Thea S. Klapwald that this perceptualevaluation method under cognitive load may be a (Department of Psychology,Barnard College,3009 Broadway,New sensitivemethod that can differentiateperceptual differences in speech York, NY 10027), and Philip E. Rubin (HaskinsLaboratories, 270 qualitybetween coded speech, even for meaningfulsentences. Crown Street, New Haven, CT 06510)
A setof perceptualstudies was conducted to assessthe salleneeof segmentalphonetic differencesbetween spontaneousand prepared FF29. Perception of /s/-eousonant clusters. Anna K. Nfib•lek, speech.Prior studiesemploying our perceptualparadigm revealed that Tomasz Letowski,and FrancesTucker (Departmentof Audiologyand naivelisteners can differentiatelexically, syntactically, and thematically SpeechPathology, University of Tennessee,Knoxville, TN 37996-0740) identicalsentence pairs that have beenextracted from brief monologs ( )40 s), identifyingthe spontaneouslyuttered version at high levelsof Nonsensesyllables/s/C/at/, whereC was/p,t,k,f,m,n,l,w/, werere- performance.No singleacoustic attribute appears to predictthe percep- cordedthen degraded either by noiseor by reverberation.The goalwas to tualsalience of spontaneityin production,though the variability off0 as comparethe effects of reverberantoverlap of high-frequencyenergy of/s/ the sentenceends and the averageduration of syllablesare both correlated with maskingby speechspectrum noise on the identificationof the con- with the detectionof spontaneity.The presentinvestigation exploits the sonants.The resultsfor l0 normal-hearingsubjects indicated that the perceptualdifferentiation test to determinethe salienceof segmentaldif- identificationof consonantscould be degradedby both noiseand rever- ferencesbetween spontaneous and prepared utterances.Against the beration.In noisethere wasa predominanceof mannererrors whereas in
$69 d. Acoust.Soc. Am. Suppl.1, Vol. 83, Spdng1988 115th Meeting:Acoustical Society of America $69 reverberationthere was a predominanceof placeerrors. It ispossible that audibleperceptual response can be maintainedduring brief periodsof the placeerrors in reverberationwere not relatedto maskingby overlap- silence.Examples of applicationof theseconcepts to a varietyof speech ping energyfrom/s/but to temporalsmearing within the consonants, sounds,such as stops and fricatives,will be illustrated.[Work supported whichobscured the offsetof the consonants.To assessif any maskingby by NINCDS and AFOSR.] energyfrom/s/could degradeidentification of followingconsonants, the samesyllables were tested with a whitenoise shaped with spectrumof/s/. In thiscondition, the consonants were highly identifiable. While energy in the consonant/s/wasrelatively high, there was little overlappingof the FF33. Codingof nrticuintory-neonstieevents in a modelof the peripheral high-frequencyspectrum of/s/with the low-frequencyspectra of tested anditory system.Z. L. Wu, P. Escudier, J. L. Schwartz, and R. Sock consonants.It is speculatedthat degradingeffects of reverberationwere (Institut de la Communication Parl•e, U.A. CNRS INPG, 46 Av. F•lix relatedto the internalsmearing within the consonants.[Work supported Viallet, 38031 Grenoble CAdex,France) by NIH.] In the last years, severalinvestigators have been concernedwith speechprocessing in auditorymodels. These lines of researchwere care- fully followedand researchwas restricted to auditorydetection of a setof FF30. The role of formantrelationships in the patternprocessing of stop articulatoryand acousticevents. The strategyconsists of exploringvar- consonants.Laurie F. Garrison(Department of Psychology,SUNY at ious processingabilities lying in the model of the peripheralauditory Buffalo, Amherst, NY 14260) systemdeveloped (see Dolmazon and Boulogne, Speech Commun. 1, 55- Interauraltransfer in selectiveadaptation was used to explorepattern 73), andattempting to demonstratea possibleadequacy between some of processingin stopconsonants. Specifically, the roleof vowelcategory as the physiologicalabilities and the acousticalstructure of speechinduced proposedby Klatt [in Perceptionand Production of FluentSpeech, edited by articulatoryconstraints. Here, first resultsabout characterizationof by R. A. Cole (Erlbaum,Hillsdale, NJ, 1980)] wasinvestigated. Here, voicingonsets and offsets and frication onsets and offsets for CVCV utter- the onset characteristics of different consonants arc "attached" to differ- ancesin French will be presentedand resultsobtained by variouscombi- cat vowel categoriesconsisting of front, back, rounded,and unfounded nationsof physiologicalmechamsms will be compared:with and without vowds.This experimenttests for patternprocessing based on theseas- neural adaptation,with enhancementor diminutionof low-frequency sumptions.The relationshipsbetween formants as well as vowel category characteristicsby temporallyor geographicallybased filtering processes. for CV syllableswere systematicallymanipulated. Five adaptorswere constructedby changing the relationships between F 1and F 2. Twoadapt- ors sharedthe samefront vowelpercept [bae]. Two other front vowel syllables,[bE] and [bI], and a backvowel syllabic, [ba], werealso used. FF34. Enhancedamplitude of the first harmonic as a correlate of Theseadaptors were tben used in an interauraltransfer adaptation experi- ment on a [bae-dae] series.The resultswill be discussedin relation to voicelessnessin aspiratedconsonants. Carol Chapin P,_ingo(Research Laboratory of Electronics,Room 36-511, MassachusettsInstitute of how changesin formantrelationships and differentvowel categories af- Technology,Cambridge, MA 02139) fect pattern processingof speechsyllables. [Work supportedby NINCDS. ] Acousticanalysis was undertaken to testthe hypothesis that informa- tionsignaling voicelessness era precedingaspirated consonant may reside in spectralcharacteristics associated with "breathyvoice" at the onsetof FF31.Duration effects on vowel perception. J. Besing,M.J. Collins,and thefollowing vowel. The studyfocused upon the enhancement of thefirst J. Cullen (Division of Communication Disorders, Louisiana State harmonic(H 1) resultingfrom an increasein the openquotient of the University,Baton Rouge, LA 70803) voicingsource waveform. The relativeamplitudes of H! and H2 were measuredin vowelsin CVd syllablespairing the consonants Theeffect of temporaldifferences on the identification of voweltokens /p,t,k,h,b,d,g/withthe vowels/i,o,u,e,A/. The syllableswere spoken by wasstudied using ten normal-hearing subjects. Stimuli were synthetically threemale native speakers of English.Harmonic amplitudes were mea- generatedand varied in durationof the steadystate, durations of the suredfrom Dl:rr spectra(computed without preemphasis) of thefirst few initial and final transitions,and F 1, F2, and F3 location. Listenerswere pitchperiods of the vowel.In general,the amplitudeof H 1 wasgreater requiredto identifythe given vowel token from a setof tenpossible alter- than that of H 2 for vowelsfollowing voiceless consonants; the converse natives.The resultssupported the establishedeffects of durationon the wastrue for voicedconsonants. A perceptualstudy using synthetic con- perceptionof vowels;however, the durationeffect was sensitiveto the tinua (varyingin VOT) is beingconducted; the continuadiffer in relative effects of initial and final transition duration and F2 location. These re- amplitudeof H 1 andH 2. Preliminaryresults indicate that for somesub- sults indicate that alterations in the duration of the initial transition has jeersan enhanced H 1may contribute to a voicelesspercept: The boundary greaterimpact than the final transitions. These effects as well as the effects betweenvoiced and voiceless stops was shifted to smallerVOT valuesfor of interactionsbetween steady-state duration, initial and final transition stimuliin whichthe amplitude ofH I wasgreater than that ofH 2 at vowel duration,and F 1,F2, F3 locationwill be discussed. [ Research supported onset.[Work supportedby NIH. ] by NIMH. ]
FF3$. An information-theOreticanalysis of aeonstic-phoneticand FF32. Sensory-perceptualtransformations in speechanalysis. James perceptualfeatures. Rick L. Jenison(Project Phoenix of Madison,Inc., D. Miller, StevenJ. Sadoff,and Mark R. Vekslcr (Central Institute for 2001 S. StoughtonRoad, Madison,WI 53716) the Deaf, 818 S. Euclid Avenue, St. Louis, Me 63110) An information-theoreticapproach using vector quantization (VQ) is In Miller's auditory-perceptualtheory of phoneticrecognition, the presentedas a frameworkfor quantifyingacoustic-phonetic distortion. sensorycff•ta of acqucncc•of glottal-sourccspectra, burst-friction spcc. An optimal codeBookderivecl from 21 C¾ monosyllablesfrom two speak. tra, andsilences are integrated into a unitaryperceptual response by the erswas generated for eachof a contrastingset of linguisticfeatures. The sensory-perceptualtransformations. The mathematicaland computa- setof feature-basedcodebooks served as distortion minimized templates tional implementationsof the hypothesizedtransformations will be de- by whichan averagedistortion could be computed for contrastingfeature scribed.One transformation is appliedto spectralpattern or shapeand is CV databasesin quietand in noise.The discriminatingpower of the fea- implementedformant by formant.For example,the center frequency of a ture codebooksbased on a mean-square-errordistortion was approxi- perceptualformant is calculatedby second-orderdifference equations mately90 percent.Consonant confusion matrices were obtained from the from thecenter frequencies, loudness, and bandwidths of the correspond- CV databasein quiet and in noisefor five subjects.Transmitted informa- ing sensoryformants of glottal-sourceand burst-frictionspectra, when tion was computedfor the samelinguistic features analyzed by vector eitheror bothare present, and from similar values for a neutralvocal tract quantization.Preliminary evaluation of featuretransmitted information duringsilence. The loudnessof theperceptual response, which is calculat- comparedto computedVQ distortionindicates a strong covariance. ed as the integratedloudness of the burst-frictionand glottal-sourcein- Further resultsand implicationsfor distinctivefeature systems will be puts,decays slowly after the cessation of sensory input and, in thisway, an discussed.
870 J. Acoust.Sec. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S70 THURSDAY MORNING, 19 MAY 1988 WEST BALLROOM A, 8:25 TO 11:40A.M.
SessionGG. PhysicalAcoustics V and Bioresponseto VibrationII: ExtracorporealShock Wave LithotripsymPhysicalAspects
Lawrence A. Crum, Chairman NationalCenter for PhysicsAcoustics, P.O. Box 847, University,Mississippi 38677
Chairman's Introduction4:25
Invited Papers
8:•0
GGI. Physicalaspects of lithotripsy.David T. Blackstock(Applied Research Laboratories and Mechanical EngineeringDepartment, University of Texas at Austin, P.O. Box 8029, Austin, TX 78713-8029and RochesterCenter for BiomedicalUltrasound, University of Rochester,Rochester, NY 14627) Lithotripsyis a medicalprocedure by whichfocused shock waves are used to breakup kidney stones. In the mostwidely used device, the Dornier lithotripter, the shockwave is producedby an electricspark at the near focusof an ellipsoidalreflector, which is locatedin a waterbath. The patient is positionedso that the kidney stonelies at the far focusof the ellipsoid.The pressuresignal there is a positivespike (headed by a shock)of amplitudeas high as 1300atto and duration about 1 #s, followedby a negativepressure of longerduration and magnitudeof order100 atto. Particular emphasis in thistalk ison thefocused sound field, which is madeup of the convergingwave reflected from the ellipsoidalsurface and a diffractedwave generated at the edgeof the reflector.The changingrelation between the two wavesalong the axis of theellipsoid gives the lithotripterpulse its characteristicwaveform. Also discussed briefly are the sparksource, pressure signal measurement, cavita- tion,and stone breakup. [Work supportedin part by NIH. ]
9:05
GG2. A comparisonof commercialextracorporeal shock wave lithotripters based on measurementsin the acousticfield. AndrewJ. Colemanand JohnE. Saunders(Medical PhysicsDepartment, Saint Thomas' Hospital,London SEI 7EH, England) A surveyof five,presently available, commercial extracorporeal shock wave lithotripters is reported.As part of the survey,a detailedexamination of the acousticfield of the Dornier HM3, Wolf Piezolith2200, TechnomedSonolith 2000, EDAP LT-01, and SiemensLithostar was carried out usingthe same25-gin, bi- laminarPVDF membranehydrophone. The absolutepressure has been estimated from the voltage waveforms (recordedon a digitalscope with a 30-MHz single-shotbandwidth) using the 1-MHz calibrationfactor for the hydrophone.The peakpositive focal pressure at clinicallyused power settings of the differentlithotripters is foundto vary in the range20-115 MPa. The peaknegative focal pressure ranges from 3-10 MPa. The width of the focalregion (defined as the regionin whichthe peakpositive pressure may begreater than 50% of that at thefocus) varies from 0.1-2.5 cm.The risetime of themain pressure peak (defined as the time from 10%-90% of the peak positivepressure) is foundto vary from lessthan 30 ns to 500 ns and the full width at a half- maximumof thepositive pressure half-cycle from 250-500 ns. The estimated peak positive pressure gain of the differentsystems range from 10 to above1000. Some of the parametersreported can be relatedto the relative clinicalperformance of thelithotripters. [Thanks to thecompanies and hospitals who provided the facilities for this survey.]
9:40
GG3. High-intensity,focused ultrasonic fields. Leif Bj•rne IndustrialAcoustics Laboratory, Technical Universityof Denmark,Building 425, DK-2800 Lyngby,Denmark) Theuse of extracorporealshock wave lithotripsy (ESWL) for disintegrationof bodystones has increased considerablyduring recent years. A worldwideactivity in thisfield is reflected in a growingnumber of interna- tionalpublications and in thedevelopment and manufacturing of several ESWL machines marketed by com- paniesin Germanyand France, in particular.Two main types of ESWL systems are prevailing, the spark gap- basedand the piezoelectricdisk-based systems. This paperis introducedby a brief reconsiderationof the featuresof pressurewaves in waterproduced by an electricaldischarge across a sparkgap emphasizing the parametersof significancetopressure wave amplitude and time course. Moreover, a comparisonispresented of resultsobtained using various theoretical models--including the KZK equation--forthe calculation of pres- surewaveform parameters, spatial pressure distribution, etc. involving nonlinearity, diffraction, and absorp- tionin thehigh-intensity focused ultrasonic fields produced by an ellipsoidas well as a sphericalcap focusing
$71 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S71 geometry.Data fromthe developmentof an ESWLof the piezoelectricdisk type are reportedincluding demandsto transducersand to the electricaldischarge circuit. Measured pressure waves in variousregions of propagationwill bepresented. Calibration demands to transducersused for finite-amplitudepressure wave detectionincluding linearity, dynamic range, etc. are emphasized. Finally, a criticalassessment is given of the mostessential pressure wave parameters, their relation to thefocal system data, and their significance to stone disintegrationefficiency.
10:lS
GG4. Extracorporeal shock wave !ithotripsy: Method, application, and physical foundation. Michael Grfinewald, Heribert Koch, and Hajo Herineking (Dornier Modizintochnik,lndustriestr. 15, D-8034 Germering, West Germany} During the lastyears a newdevice has been put into the medicalhealth system allowing for noninvasive treatmentof kidneyand gall stones: the Dornier lithotripter. The principlemethods of shockwave generation arediscussed, such as underwater spark discharge, dielectric laser breakdown, electromagnetic and piezoelec- tric method,and their suitabilityto disintegratebody concrements effectively. For practicalapplications, carefulchoice of system parameters has to be made to fulfill all requirements:good stone disintegration and low painand sideeffects for the patientssimultaneously. Based on the theoryof focusingthe relevantphysical effectslike diffraction,nonlinear propagation, and attenuation(absorption) are explored both qualitatively andquantitatively, so far aspossible. It will beshown that a full time-dependentprofile calculation can be given withinsimple diffraction theory accounting for a properdescription of the dependenceof focusingon aperture, includingstress and tensile.The relevanceof pulseshape forming will be explored.Additionally nonlinear propagationin relationto dampingand pulse deformation in softtissue is important. Within a simplemodel the propagationof highlyunharmonic pulses will be calculatedincluding nonlinear steepening, anomalous tissue absorption,and dispersion.
10:50
GGS. Cavitation microjets as a contributing mechanismfor stone disintegrationin ESWL. Lawrence A. Crum (National Centerfor PhysicalAcoustics, P.O. Box 847, University,MS 38677) The intenseshock waves that are usedfor stonedisintegration in ESWL can readily produceacoustic cavitationin a varietyof liquids.It hasbeen discovered that the damageto metalobjects placed at the focusof an ESWL can be attributedto the actionof liquidmicrojets produced by collapsingcavitation bubbles [A. J. Colemaneta!., UltrasoundMed. Biol. 1:3,69-76 (1987) ]. Furthermore,Delius hasshown that theseshock wavescan produce cavitation in oioo[Miller eta!., L Acoust.Soc. Am. $uppLI 83, S 89 (1988)]. In this paper,a generalreview will be givenof themechanism of microjetformation in cavitationbubbles and argu- mentsmade that cavitationplays an importantrole in stonedisintegration. A film will be shownof microjet formationin collapsingcavitation bubbles induced at low acousticfrequencies. [ Work supportedin part by NIH and ONR.]
ContributedPaper
11:25 pressurewave drives the Oilmore-Akuliehev formulation for bubbledy- namics;the zero order effect of gasdiffusion on bubble response isinclud- GC•. A theoreticalstudy of cavitafiongenerated by an extracorporeal ed. Resultspredict that a preexistingbubble in the micronrange will shockwave !ithotripter.Charles C. Church (Center for Biomedical expandto over100 times its initial size, Ra, for 240/ts,with a peakradius Ultrasound,319 Hopeman Hall, TheUniversity of Rochester,Rochester, of 1350Ro,then collapse very violently,emitting far UV or soft x-ray NY 14627) photons(black body assumption). Gas diffusiondoes not appreciably mitigatethe violenceof the primarycollapse, but doessignificantly in- The intenseacoustic wave generated at the focusof an extraco•l creaseRo, up to 40/tin, as well as the durationof subsequentringing, shockwave!ithotripter is modeledas the impulseresponse of a parallel assumingno break up or productionof microbubbles.The predictionof x- RLC circuit.The shockwave consists of a zero rise time positivespike rayemission is of moreacademic than clinical concern because relatively whichfalls to 0 at 1/•s followedby a negativepressure component 6 ps fewphotons will beproduced, and renal tissue exhibits a fairlylow radio- longwith amplitudes scaled to + 1000and -- 167bars, respectively. This sensitivity.[Work supportedby NIH. ]
$72 J.Acoust. Soc. Am. Suppl. 1, Vol. 83, Spdng 1988 115thMeeting: Acoustical Society of America S72 THURSDAYMORNING, 19 MAY 1988 GRANDBALLROOM A, 8:30A.M. TO 12:05P.M.
SessionHH. Musical Acoustics III: Arthur H. BenadeMemorial Session I: Horns, Strings, and Harmony
Carleen M. Hutchins, Chairman CatgutAcoustical Society, 112 Essex Avenue, Montclair, New Jersey 07042
Chairman's Introduction---8:30
Invited Papers
8:35 HH1. Arthur H. Benade:The makingof a musicalacoustics researcher, theorist, and teacher. Virginia Benade(3126 Woodbury Road, Cleveland, OH 44120) TheRampur State Band, childhood consumption, andrepairing World War II aircraftradios are only a few of theinfluences reported on here as having had some bearing on A. H. Benade'sdevelopment asa musical acoustician.
9:05
HH2. Arthur Benade•Synthesizer.George Jameson (6214 38thStreet, Kenosha, WI 53142) Ofthe talents that led to Art Bcnade'sconsiderable accomplishments in musical acoustics, one has escaped prominentmention--namely--his proclivity to synthesize.He hada strongability to select ostensibly unrelat- edfacts and thoughts from many disciplines, theoretical and practical, over a periodof time. When these stored itemsbegan to fit together,Art was often able to discerna directionand reach the solutionto an acoustical problem.The breadth of overallknowledge required to permitaccess to obscurerelationships has often been itselfobscured by the impact of Art'sresults. He calledfreely on branches of physics,chemistry, mathematics, andelectronics not centralto hiswork. Besides the sciences,he waswell grounded in music,literature, and manyphases of engineering.On a differentlevel, he wasa proficientmachinist, mechanic, and tinkerer. Not least,he was a usablemusical performer. He possessed a truly remarkable musical ear that was a majorsource of information.While many of theabove abilities are requisite for anysuccessful experimental physicist, Art extendedhis attentions further--to the user of musicalintruments. Examples of thesynthetic process will be cited--partlyfrom the standpoint of theauthor, whose background is in manufactureand repair of instru- ments,and in routineprofessional performance and teaching of woodwinds.
9:35
HH3. Fundamentalsof pianoscale design. Earle L. Kent (2510 RiverviewPlace, Elkhart, IN 46516) Thedetermination of optimum piano string diameters, speaking lengths, and total lengths is complicated dueto theinteraction between the strings, their breaking strengths, available space for them, the energy they canaccept from hammers, and the utilization of string energy to producesound. Many economic, philosophi- cal,and psychological factors in additionto the physical considerations areinvolved. No universallyaccepted definitionof good piano tones has been established. Though it isnot practical to lay down a setof specific rules forpiano scale design, it ishelpful to considerthe relationships between involved parameters in order to guide thedesign, rebuilding, and repair of pianos. A setof string design relationships isgiven to serve as a benchmark in establishingpiano string dimensions and as a basisfor making comparisons in existing pianos. It seems reasonableto consider starting with well chosen string dimensions since the strings are the starting sources of pianotones.
10:05 HH4. Fromtwenty years of violinacoustics research. Erik V. Jansson(Department of Speech Communicationand Music Acoustics, Royal Institute of Technology(KTH), P.O.Box 700 14,S-100 44 Stockholm,Sweden) A veryfruitful year as research associate to Arthur Benade in 1967started my professional work as a scientist.This year formed much of my way of thinking and future work on strings, some of the "highlights" of whichare reviewed. Normal modes ofplates ofa violin under construction weremapped bymeans ofhologram interferometry(with Molin and Sundin). Experiments with the air cavity, employing Fransson's Ionophone, disclosedalarge number ofair volume modes. Long-time-average spectraof quality judged violins implied that highlevel below 1kHz, low at 1.3kHz, and high around 2.5 kHz are favorable (with Sundberg and Gabriel- son). Normalmodes of thecomplete violin were mapped with "speckle interferometry" andinput admittance measurements,andimplied that "equally strong" low-frequency modes and a marked"bridge hill" at 2.5kHz
S73 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S7'3 arepositive for quality (with AlonsoMoral). Theinfluence of material properties and design of freeplates were investigatedin physicalexperiments and finiteelement calculations indicating that the thicknessis the main factor(with JakubNiewczyk, Molin, and Lindgren).The propertiesof platesin a simplifiedviolin body are presentlyinvestigated (with BenedyktNiewczyk). I believethat the learningsfrom my year with Arthur Benade,thus, have further been spread during the last 20 yearsto my fellowcolleagues in Swedenand guest researchersfrom Spain,Poland, and Australia.
10:35
HHS. Listeningto soundin rooms.W. M. Hartmann(Physics Department, Michigan State University, East Lansing,MI 48824) ProfessorA. H. Benadefrequently remarked on theparadox involved in listeningto musicalsounds in rooms.He notedthat thetransmission path from the sound source to the listeneris unreliable;its impulse responseand its correspondingfrequency response, are irregular, unpredictable, and unstable.Nevertheless, the listenerobtains reliable information about the locationof the source,about the tone color, and aboutthe transientcharacter. The paradox may be resolved in principleby postulating that the listener's nervous system performscertain temporal summation and differencing operations as well as binaural summation and differ- encingoperations on theinternal representation of the sound field. The present talk will reviewwhat is known frompsychoacoustical experiments concerning these parallel summation and diffcrencing operations with the goalof distinguishingbetween plausible and implausible models. Most striking has been recent progress on binauralprocessing where psychoacoustics andelectrophysiology appear to beconverging on a singlemodel, basedupon crosscorrelation in tuned channels. Of particularimportance is the precedence effect, which, after 100 yearsof study,still createssurprises but still wantsan adequatedefinition. [Work supportedby the National Institutesof Health. ]
11:05
HH6. Art Beuadeat theInstitut de Rechercheet CoordinationAcoustique/Musique in Paris. Ren6 Caussc (IRCAM, 31, rue Saint-Mcrri,75004 Paris, France) Thispresentation dicusses Art Benade'sinfluence on different instrumental acoustics projects at IRCAM, a musicalresearch institution headed by composerPierre Boulez. Art Benadewas a memberof IRCAM's originalscientific advisory committee. New performance techniques, such as flute and trombone multiphonics, inputimpedance measurements of wind instruments and mutes for brasses were among the projects in which Art Benademade contributions, asa theoreticalphysicist and an adept experimenter, but also as an experi- encedmusical performer and instrumentmaker. Thesemultifaceted talents enabled him to communicate effectivelywith composersand performers,to understandmusical problems in physicalterms (and vice- versa),and to participatein everystep of theseprojects. We will presentthe most recent results of someof these projectsand discussmore particularly a newmodel of the "wah-wah"mute for the frenchhorn.
11:35
HH7. Remini•encesof Art Benade.Harry B. Miller (Code2131, Naval Underwater Systems Center, New London, CT 06320) In theearly 1950's, Art Bcnadegave a two-semesterevening course on atomic physics. I was a memberof theclass. We soon got into after-class discussions of applied acoustics and these never really stopped until his death. Some of this will be recounted.
THURSDAY MORNING, 19 MAY 1988 GRAND BALLROOM B, 8:30 TO 11:15 A.M.
SessionII. PsychologicalAcoustics V: Psychoelectronicsand Psychoacousticsin ImpairedVersus Normal Listeners
David A. Nelson, Chairman HearingResearch Laboratory, University of Minnesota,2630 UniversityAvenue, S.E., Minneapolis,Minnesota 55414
ContributedPapers
8:30 Gapdiscrimination was measured as a functionof standard-gapdura- tion and markerlevel in both normal-hearinglisteners and in patients II1. Gapdiscrimination in normallisteners and enchlear implant patients. with Nucleusand Symbioncoehlear implants. For normallisteners, the RobertV. Shannonand Donna L. Neff (BoysTown NationalInstitute, standardgap ranged from 0 to 400 ms and the broadband-noisemarkers 555 N. 30th Street, Omaha, NE 68131) were30 to 90 dB SPLtotal power. For theimplant patients, the range of
S74 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America $74 standardgaps and markerlevels was similar, but the markerswere either onewere added in cosinephase. The remainingharmonic started in cosine sinusoldsor pulsetrains. Normal listeners were tested under both motmu- phase,but was shiftedin phasehalfway through either the first or the ral anddichotic (one marker to eachear) conditions.When expressed as secondof thetwo stimuli used in a 2-AFC trial.The taskof thesubject was AT/T, the gap discriminationfunctions for both normalsand implant to identify the stimuluscontaining the phase-shiftedcomponent. An patientsare nonmonotonic with a peakat standarddurations of 10-20ms. adaptiveprocedure was used to meaureDLs for phasefor threenormally Performanceat the shorteststandard durations is markedlypoorer with hearingsubjects and four subjects with unilateralcochlear impairments. dieboricpresentation. The nonmonotonicityappears to reflecta change Whenthe normalears were tested at a levelof 70 dB SPL percomponent, froma tasklimited by peripheralprocesses to onelimited by morecentral Dis were smallest(2'-4') for harmonicsabove the eighthand for the processes.Cochlear implant patients exhibit a similarchange. [Work sup- lowestfundamental frequency (F0). Changesin phasecould not be de- portedby NIH. ] tected for harmonic numbers below four at the lowest F0 and below 8-13 at the highestF0. The DIS increasedmarkedly with decreasinglevel, exceptfor the highestharmonic, where only a smalleffect of levelwas observed.The Dis for theimpaired ears were generally larger than those
8:45 for the normalears when the comparison was made at equalSPL but the discrepancywas reduced when the comparisonwas made at equalSL. For 11:2.Aging effects on gapdetection thresholds amon$ normal and hearing low harmonic numbers, a few caseswere found where DIS were smaller impairedsubjects. Stanley A. Gelland (Departmentof Commonicafion for the impairedthan for the normalear of a givensubject, even at equal Arts andSciences, Queens College, Flushing, NY 11367),Iody Porra.zzo SPL. This effectmay be explicable in termsof the broaderauditory filters (City Universityof New York, New York, 10036), and ShlomoSilman of theimpaired ears. The resultswill becompared to measuresof temporal (BrooklynCollege, Brooklyn, NY 11210) resolutionand pitchdiscrimination using the samesubjects. [Work sup- portedby the Medical ResearchCouncil, U.K.] Age differencesbetween normal and hearingimpaired groups have cloudedthe effects of hearingloss on temporalselectivity in manystudies. This studyattempted to addressthis problem by comparinggap detection thresholds(GDTs) betweena normal hearinggroup and a groupwith primarilyhigh-frequency sensorineural hearing losses, each containing subjectswith similarage ranges.Transformed up-down estimatesof 9:15 70.7% GDTs wereused in a two-alternativeforced-choice paradigm with feedback.Each run involvedsimultaneous estimates of two gap thresh- 114.Intensity discrimination at threefrequeneie• by meansof pulsed-and olds, using interleavedup-down procedureswith random switching continuous-toneparadigms. C. W. Turner, I. I. Zwislocki, and P. betweenthem. Gap detectionthresholds varied from approximately 3 to R. Filion (CommunicationSciences and Disorders and Institute for 16 ms amongsubjects in both groups.Across all subjects,GDTs were SensoryResearch, Syracuse University, Syracuse, NY i 3244-2280) significantlycorrelated with ageand with hearingsensitivity at 12 003 Hz Differencelimens (DL's} for the intensityof puretones were mea- but not with auditorythresholds at any other frequencies.Regression suredwith 2AFC pulsed-toneand continuous-toneparadigms at 500, analysesrevealed that age was the only significantpredictor of GDTs 2000,and 6000 Hz acrossa broadrange of sensationlevels (SL's). In the amongthese subjects. The findingssuggest that temporalprocessing is pulsed-toneexperiments, the pedestalwas gated simultanenusly with the affectedby age-related auditory changes that are not necessarily reflected increments,whereas, in thecontinuous-tone experiments the pedestal be- as shiftsof absolutesensitivity. gan400 msbefore the firstobservation interval, and continued through- out the trial until 400 ms after the termination of the second observation interval.The DL's obtainedon five normalsubjects were lower for the continuous-paradigmthan for the pulsed-toneconditions for all testfre- 9.'OO quenciesand SL's. The differenceagrees with physiologicalresults show- II3. Differencelimens (DLa) for phasein normallyhearing subjects and ing that thesingle-unit neural response to an incrementis independent of subjectswith unilteral •.hlear hearing loss. Brian C. J. Moore and whetherthe increment is presentedto fibersthat are adapted (continous- Brian R. Ginsberg (Department of Experimental Psychology, toneexperiment) or unadapted(pulsed-tone) experiment [Smith and Zwislocki, Bioi. Cyber. (1975)] and also with the argumem that the University of Cambridge, Downing Street, Cambridge CB2 3EB, England) pulsed-toneparadigm involves the additionalrequi•ment of a "memory load" for subjects.However, according to our preliminaryfindings, the Complextones containing the first 20 harmonicsof 50, 100,or 200Hz, differencedoes not hold for all hearing-impairedsubjects. [Work sup- all at equalamplitude, were used as stimuli. All of theharmonics except portedby NIH Grant NS24255.]
Break
9:30-9-.45
9'.45 thoseof normal-hearinglisteners, i.e., hearing-impaired listeners did not demonstrateabnormal upward spread of maskingwhen equivalent mask- er levelswere compared. Only 8 of the23 hearing-impairedlisteners dem- I15. Frequencyresolution at equivalentsound-pressure levels in normal- onstratedabnormally broad PTCs, due exclusively to abnormallygradual hearingand hearing-impairedlisteners. David A. Nelson (Hearing high-frequencyslopes to their PTCs. This demonstrationof abnormal ResearchLaboratory, University of Minnesota,2630 UniversityAvenue downwardspread of maskingwas only observed in listenerswith hearing S.E. Minneapolis,MN 55414) lossesgreater than 40 dB HL. From theseresults, it would appearthat some,but not all, cooblearhearing losses greater than 40 dB HL influence Forward-maskedpsychophysical tuning curves (PTCs) were ob- the sharptuning capabilities usually associated with outerhair-cell func- tained for multiple-level 1000-Hz probe tones from 15 normal-hearing tion. The restrictionof abnormalfindings in this study to the high-fre- and 23 hearing-impairedears with cochlearhearing loss. Comparisons quencysides of PTCs aloneis likely due to the useof equivalentmasker- betweennormal-hearing and hearing-impaired PTC characteristicswere levelcomparisons, and alsoto the useof flat-envelopenonsimultaneous madeat equivalentmusket levels near the tipsof PTCs.The low-frequen- maskersrather than fluctuating-envelopesimultaneous masker. [Work cy slopesof PTCsfrom hearing-impaired listeners were not different from supportedby NINCDS. ]
S75 d. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S75 10:00 stimulifor eachsubject were selected on the basisof predictionsfrom Articulation Index theory, in order to control for thresholdsensitivity 116. Forward-maskingproperties of multicomponentsignals in normal differencesamong the normal-hearing and hearing-impaired listeners. Es- and hearing-impairedsubjects. Munah Kim, ChristopherW. Turner, timatesof filterwidth for 25 hearing-impairedsubjects (calculated from and Evan M. Relkin (Communication Sciences and Disorders and masked thresholds measured in broadband noise as function of the width Institute for SensoryResearch, SyracuseUniversity, 805 S. Crouse of a spectralnotch) revealed auditory filter widths that increased with the Avenue, SyracuseNY 13244-2280) thresholdfor the probe.Consonant place perception for most hearing- The forward-maskingproperties of inharmoniccomplex stimuli were impairedsubjects was nearly equivalent to resultsfor normal-hearinglis- measuredboth for normaland hearing-impaired subjects. The signalwas tenerswhen thresholddifferences were controlled.These results suggest a singlesinusold of 1000Hz. Signalthreshold was obtained for sixdiffer- that frequencyresolution and speech recognition may coexist as indepen- ent maskersthat variedin the numberof components.Masking stimuli dent consequencesof thresholdelevation. [Work supportedby grants consistedof 1, 3, 5, 7, 9, or 11 components,logarithmically spaced in from NINCDS and the DeafnessResearch Foundation. ] frequencysurrounding the signal, and wcrc presented at a fixedlevel of 80 dB SPL percomponent. An adaptivetwo-alternative forced choice meth- od was used to determinethresholds. In normal-hearingsubjects, the thresholdfor the signaldecreased as the numberof componentswas in- creased,demonstrating that signalswith morecomponents are lessaffec- tire maskers.Results from hearing-impairedsubjects indicated no de- 10:45 creasein thresholdwith increasingmasking components, instead, the thresholdsincreased as more components were added to the firstmasker. If9. Auditory temporal summationin normal-hearingand hearing- [Work supportedby the DeafnessResearch Foundation and the Whi- impairedcat. JanetM. Soleckiand GeorgeM. Gerken (Cailler Center taker Foundation.] for CommunicationDisorders, Uiversity of Texasat Dallas, 1966Inwood Road, Dallas, TX 75235)
Auditory temporalsummation functions were measured for five cats beforeand after they were exposed to a 2-kHz toneat 110dBA for 48 h. Ten digitallygenerated stimuli with overalldurations ranging from 8.32 10:15 to 275 mswere used. Stimulus frequency was 6.25 kHz. Six stimulicon- sistedof single tone bursts and the remaining four contained multiple tone 117. Growth of masking in listeners with cochlear impairmentsand bursts.Twelve acceptablethresholds for each stimuluswere obtained impairmentssimulated by masking.S•Jren Buus (Communicationand from eachanimal before and after exposure.Threshold measurements Digital Signal ProcessingCenter, 409 DA, NortheasternUniversity, beganagain no sooner than 27 dayspostexposure. Prc- and postexposure Boston,MA 02115) and Mary Florentine (CommunicationResearch audiogramswere also obtained and meanpermanent threshold shift at Laboratory, 133 FR, NortheasternUniversity, Boston,MA 02115) 6.25kHz was32.6 dB. Pre-and postexposure temporal summation func- tionswere characterized by theslope of the regressionlines for threshold The higher-than-normalmasked thresholds observed in many im- versuslog duration.The meanslope in dB per decadeof durationwas pairedlisteners may be explainedby nonlinearadditivity of maskingby -- 6.6 preexposureand - 3.8 postexposure.The slopefor normal-hear- the externalmasker and an hypotheticalinternal masker [Jesteadt et al., ingcat is very similar to theslope of - 6.1 thatwas obtained in a compan- J. Acoust.Soc. Am. Suppl.1 81, S77( 1987);Humes etal., J. Acoust.Soc. ion studythat usednormal-hearing human subjects. Am. Suppl. I 81, S77 (1987)]. To testthis hypothesis,Btktsy tracking was usedto measuregrowth-of-masking functions for 250-mstones with frequenciesat and abovethose of continuouslow-pass or octave-band noisemaskers. Results for four impairedlisteners show that the mini- mum-effective-maskerlevels (MEMLs, abovewhich maskingoccurs) arehigher than normal for smallmasker-signal separations, Af's, but ap- 11:00 proachnormal values for largerAf's. Abovethe MEMLs, maskinggrows 1dB/dB or less.Results for simulatedimpairments are similar to thosefor II10. Some parameters of automatic audiometric thresholds. I. realimpairments, but MEMLs tendto behigher and masking often is less M. Young and L. D. Lowry (Department of Otolaryngology,Jefferson in simulatedthan in realimpairments. These results indicate that nonlin- Medical College of Thomas JeffersonUniversity, Philadelphia,PA ear additivityof internaland external masking accounts for some,but not 19107) all, of the excessivespread of maskingobserved in manyimpaired listen- ers. [Work supportedby NIH.] Thresholdmeasurements wcrc made by two automaticaudiometers; ( 1) continuouslyvariable frequency tracing audiometer with an attenu- ationstep of 0.25 dB; (2) continuouslychanging discrete individual fre- quency tracing audiometer with an attenuation step of 1.0 riB. Five trainedsubjects with bilateralnormal hearing, four subjectswith unila- teral cochlearlesion, and four subjectswith unilateralretrocochlear lesion weretested. For both pulsedand continuoustones, there was no observ- 10:30 able difference in thresholds between two audiometric measurements for IIS. Auditory filter width andconsonant recognition for hearing-impaired subjectswith bilateralnormal hearing and for carswith normalhearing in listeners.Judy R. Dubno and Donald D. Dirks (Division of Head and subjectswith unilateralhearing impairment. The earswith cochlearlesion Neck Surgery,UCLA Schoolof Medicine, Los Angeles,CA 90024-1794) indicatingabnormal rapid adaptationshowed slightly greater separation betweenpulsed and continuous tone thresholdsby continuously variable Abnormal frequencyresolution is often assumedto contributeto re- frequencytracings than by continuouslychanging discrete frequency duced speechrecognition by hearing-impairedlisteners. However, a tracings.For the carswith retrocochlearlesion indicating abnormal slow strongassociation between these two conditionssecondary to hearing type of adaptation,much greater separation was demonstratedby the impairmenthas not beenestablished, perhaps because both are degraded former. Thesefindings were discussedwith and relatedto parameters by thresholdelevation. Thus, in this experiment,which examined audi- affectingthresholds; slow attenuation step in size,threshold at equilibri- tory filter effectson consonantrecognition, presentation levels of speech um (stabilizedthreshold), and lateralfrequency spread of adaptation.
$76 J. Acoust.Soc. Am. Suppl. 1, Vol. 83, Spring 1988 115th Meeting:Acoustical Society of America $76 THURSDAY MORNING, 19 MAY 1988 EAST BALLROOM A, 8:30 TO 10:30A.M.
Session JJ. Structural Acoustics and Vibration III: Plates and Shells and Radiation Problems
SabihI. Hayek, Chairman EngineeringScience and Mathematic• Applied Research Laboratory, The Pennsylvania State University, University Park, Pennsylvania16802
ContributedPapers
8'3O A modelto solvecoupled fluid--structure interaction problems, such as acousticscattering from an elasticbody, is described.This modeluses JJl. An analytic model for vibration of turbomachineryrotor blades. doublycurved, nine-noded, isoparametric finite elements over which the JamesP. Charpieand CourtneyB. Burroughs(Applied Research pressureis takento vary quadratically.Results from thismodel are com- Laboratory,Pennsylvania State University, P.O. Box 30, StateCollege, paredwith anothercomputer model (NASHUA) that is basedon planar PA 16804) finite elementsover which the acousticpressure is assumedto remain An analyticmodel for thevibration response of turbomachineryrotor constant.Several examples are presentedto demonstratethe reductionin bladesthat includesthe effectsof variationsin blade thickness,curvature the numberof elementsneeded for accuratemodeling. Comparison of of the bladeedges, and variations in the curvatureof the blademiddle solutiontime and accuracyis alsogiven. surfacein twodimensions isdeveloped. Mathematical fits to both surfaces of theblade are generated in termsof polynomialequations with constant coeflieients.The polynomialequations for the blade surfacesare then 9:15 usedin shellequations, based on Love'sthin shellassumptions, to gener- atedifferential equations governing the vibration motion of theblade. The JJ4. Acousticalpower radiation from cylindrical shells,with or without coefficientsin the blade-fittingpolynomials act asparameters in theequa- stiffeners: Understanding of phenomena and practical aspects. tionsof motionfor the blade,so that equationsfor a newblade can be B. Laulagnet,J. L. Guyader,and C. Lesueur(Laboratoire Vibrations- developedeasily by changingthe coefficientsin the governingequations Aeoustique,Institut National des SciencesAppliqu6es de Lyon, 69621 to thosegenerated for the newblade. An approximatemethod of solution Villeurbanne Cedex, France) for the resonantresponse of a sampleblade is presented. An analyticalstudy of radiationby finite,stiflened cylindrical shells is presented.The modelof stiffenerutilized allows one to treat hollowcross- sectioncases encountered in industry.The point drivingforce is on the 8:45 stiffeneror directlyon theshell and can be radial, tangential, or longitudi- nal. Fluid and structureequations are solvedwith the modal analysis, JJ2. Modificationof Lamb wave spectrumin fluid coupledplate. S. usingthe in oacuononstiffened basis. This leadsto the calculationof L Rokhlin (Departmentof Welding Engineering,The Ohio State radiationimpedances and massand stiffnessgeneralized terms of stiffen- University,190 West 19Avenue, Columbus, OH 43210), D. E. Chimenti ers.Theoretical results are presentedboth in air and in water, in radiated {Materials Laboratory,Wright-Patterson Air Force Base,Dayton, OH power,radial quadratic velocity, and radiationfactor. Mechanical modal 45433), andA. H. Nayfeh{Department of AerospaceEng., M.L. 70, cross-couplinginfluence is shownas well asthe stiffening influence on the Universityof Cincinnati,Cincinnati, OH 45221) in water vibroacousticshell behavior.Parameters like rigidity, number It hasbeen accepted previously that for mostpractical eases when a and distributionof stiffeners,shell and stiffener structural damping, loca- plate is immersedin a liquid the Lamb wave spectrumis only slightly tion, and forcetype are studiedin air. Finally meansfor noisereducing in perturbeddue to energyloss related to leakageof the wavefrom the plate air arededuced. [ Work supportedby theMinist6re de la D6fense,Contrat to theliquid. In contrastit wasrecently discovered by D. E. Chimentiand D.R.E.T. No. 85065. ] A. H. Nayfeh [Appl. Phys.Lett. 49 (9) (1986)] that for leakyLamb wavesin a graphiteepoxy composite plate in water a stronganomaly appearsin the topologyof the So modeas a functionof frequency.Al- thoughit wascaused by the fluid,the reasonfor thisanomaly was still not 9:30 clear.The natureof the anomalousbehavior of leaky Lamb wavesis dis- cussedin thispaper. It is shownthat for an understandingof the spectrum JJS. Soundradiation from a vibratingbody in motion. Xiao-Feng Wu topology,the full complexspectrum of propagatingand nonpropagating andAdrian Akay (DepartmentofMeehaniealEngineering,WayneState modesshould be considered.It is demonstratedthat the topological University,Detroit, MI 48202) changesin the So modeare relatedto its nearnessto the first enmplex Paststudies of acousticradiation problems have been enncerned with branchof the spectrom.Increasing liquid densityleads to interaction soundwaves generated by either a vibratingbody or a movingpoint betweenthese modes and mutualinterchange between their branchesin sourcewith no dimensionsof its own.This paperpresents a methodfor the complexplane. It is shownthat, in general,as the densityof the im- determiningthe acoustic field radiated from an arbitraryobject undergo- mersionliquid increasesfrom zero to infinitythe complexspectrum of ing, simultaneously,a harmonic oscillation and a spatialmotion with platemodes gradually transforms from a spectrumof Lamb wavesin a respectto the surroundingfluid medium.At thisstage of the investiga- freeplate to a spectrumof wavesin a plateclamped on its surfaceswith tion, the turbulenceand flow distortioncaused by the motionof the vi- slip(boundary conditions ar• = 0, O'•= 0). bratingbody are neglected.The methodis basedon the assumptionthat thesurface of the vibrating body can be described in termsof a distribution of simplemonopole and dipolesources. Utilizing a linear coordinate 9.•0 transformation,one is then able to reducethe problem to thatof radiation from a stationarysource for which the analyticalsolution has beenob- JJ3. The solution of fluid-structure interaction problem using tained[ Morseand Ingard, Theoretical,4coustics (Princeton U.P., Prince- quadratic. doubly curved finite elements. Joseph J. Shirron and ton, NJ, 1986) ]. The total acousticfield is the sumof contributionsfrom Luise Schuetz(Naval ResearchLaboratory, Code 5134, 4555 Overlook all the simplesources on the surfaceof the movingobject. The methodis Avenue,Washington, DC 20375-5000) illustratedusing a vibratingsphere moving with eitherconstant or time-
S77 J. Acoust.Soc. Am. Suppl. 1, VoL 83, Spdng 1988 115th Meeting:Acoustical Society of America $77 dependentvelocity profiles relative to the surroundingmedium. Exten- oftarget material results in an impulsive reaction transmitted tothe struc- sionof thismethod to impulsivemotions is discussed.[Work supported ture. Differenttarget materials are testedto enhancethe efficiencyof by NSF andWSU Institutefor ManufacturingResearch. ] energyconversion. A clamped beam and a cantileverbeam are used in the experimentsand the measured natural frequencies are compared favor- ablywith the calculated ones. Beam splitters and multiple targets are used to testthe concept of multipointsimultaneous impulsive loading. 9:45
JJ6. Leaky Lamb wave inspectionof thin plates.Peter B. Nagy and Laszlo Adler (Department of Welding Engineering,The Ohio State University,Columbus, OH 43210)
A single-transducerimmersion technique is introducedto studydis- 10:15 persionof leakyLamb waves in differentplates, such as fiber-reinforced compositelaminates and adhesivelap joints.The plate understudy is JJS. Sound radiation control for two plates assembledin an L shape. interrogatedby a broadbandpulse from both sides when placed between PascalF. Millot (Laboratioredes Sciences de i'Habitat, EcoleNationale thetransmitting-receiving transducer and a perpendicularlyaligned mir- des Travanx Publics de l'Etat, Rue Maurice Audin, 69 120 Vaulx-en- ror. The transmittedpeaks in thespectrum of the receivedecho from this Vclin, France) and Jean-LouisGuyader (Laboratoirc Vibrations mirror correspondto doublemode-converted leaky Lamb wave reso- Aconstique,Institut National des SciencesAppliqutes, 69 621 nances.These peaks are measuredat differentangles of incidence,i.e., at Villeurbanne Cedex, France) givenphase velocities to get velocitydispersion curves. Experimental re- Reductionof soundradiated in a cavityby an L-shapestructure excit- suitsare presented to demonstratethe accuracy and versatility of thisvery ed by mechanicalforces, can--in someinstances--be achieved by taking simplemeasuring technique. advantageof the phenomenathat are uniqueto that geometricalshape. Insteadof relyingonly on the controlof the velocityof the structureto reducethe soundradiated, it will bc shownthat it is possibleto select modesof the assembledplates that havea low-radiationimpedance, by 10:00 positioningcorrectly the mechanicalexcitation forces. At low frequen- J J7. Transient structural vibration created by a laser beam. Yousef cies,improvements can reach several decades for the mean-squaredpres- A. Abo-Mossallam(Faculty of Engineering,EI-Mansoura University, sureinside tbe cavity for identicalplates, while above the limit frequency Egypt) of thecavity no improvement should be expected with this method. Below the limit frequency,improvements will greatlydepend on the respective Experimentalwork is conductedto investigatethe transientvibration dimenions of the plates,as well as on the valuesfor the first resonant of a structureby an argonlaser beam. The conversionof laserlight into frequencyof the plates and of thecavity. Theoretical results, derived from mechanicalenergy has been studied through focusing laser beam to va- a modalapproach for the cavityand the plates,as well as experimental porizematerial from a targetattached to the structure.The rapidejection resultsobtained for plasterplates in laboratory,will be presented.
THURSDAY MORNING, 19 MAY 1988 WEST BALLROOM B, 8:30 A.M. TO 12:05 P.M.
Session KK. Underwater Acoustics V: Bottom Interaction
Barry J. Uscinski, Chairman Departmentof AppliedMathematics and TheoreticalPhysics, Cambridge University, Cambridge CB3 9EW, UnitedKingdom
Chairman's Introduction--8:30
ContributedPapers
8:35 edpreviously [J. Acoust.Soc. Am. Suppl.1 82, S121(1987) ]. Thebehav- ior is alsoshown to bequalitatively similar to thatof an exactsolution for KKI. Computationof the Blot drag and virtasl m•ss coefficientsfor a the caseof straightcylinder pores obtained by Blot and other investiga- three-dimensionalpore space. B. Yavari and A. Bedford(Department tors. Wave velocityand attenuationof compressionalwaves calculated of AerospaceEngineering and EngineeringMechanics, The Universityof basedon thesecoefficients are shownto agreereasonably well with experi- Texas, Austin, TX 78712) mentaldata. [Work supportedby ONR.] Applicationof the Blottheory to the studyof the acousticwaves in oceansediments requires evaluation of the dragand virtual mass coeffi- cients.These coefficients depend on thefrequency and microstructure of the sediments.Recently, a methodwas developed to evaluatethese coeffi- 8:50 cientsby Bedfordetal. [J. Acoust.Soc. Am. 76, 1804-1809(1984) ]. The KK2. Shear wave velocity in surfictal marine sediments:A comparisonof methodrequires solving for themotion of thefluid in thepores when the in situ and laboratory measurements. Michael D. Richardson, porewalls are subjected to anoscillatory motion. The finite element meth- Enrico Muzi, and Luigi Troiano (SACLANT UnderwaterResearch odhas been used to determinethe fluid motion. The drag and virtual mass Centre, I-19026 La Spezia,Italy) coefficientshave been determinedfor a microstructurecomposed of sphericalgrains saturated with water.The qualitativebehavior of the In situand laboratory measurements of shearwave velocity were com- coefficientsis shownto be similarto the two-dimensionalresults present- paredfor a widevariety of shallow-watermarine sediment types found in
S78 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spdng1988 115th Meeting:Acoustical Society of America S78 thevicinity of La Spezia, Italy. Both types of measurements utilized pulse PacificRise [measured by Lonsdaleet al. (1980)] wascomputed versus techniques,with shear waves generated and received by bimorph ceramic averaginglength for predictionsof scatteringloss versus Fresnel zone bender elements.Measurements were made between200 and 1000 Hz. In dimensions.The source-receiver geometry replicated previously, report- situmeasurements and sediment collections were made by scuba divers to ed measurementsof the reflectioncoefficient versus grazing angle at a minimizesediment disturbance. Values of shearwave velocity were low nearby,sediment-free site [Diachok et al. (1986)]. Thecomputed rough- ( < 100m/s) forall sedimenttypes sampled. Typical values ranged from nessincreased slowly with decreasing grazing angle (O) at large0, being 20 m/s in high porosityclays to 50 m/s in sandysediments. Shear wave about8 m (at 20Hz) at 60',and increased rapidly with decreasing grazing velocitiesmeasured by in situand laboratory techniques were not signifi- anglefor small0, beingabout 38 m (at 20 Hz) at 10'along a trackthat cantlydifferent, suggesting that shear wave velocities measured on care- paralleledthe ridge axis. The effectiverms roughness inferred from the fullycollected core samples of surficialsediments need not be corrected to reflectivitydata for 50ø < O< 70'was 6 m (at 20Hz alonga paralleltrack), in situconditions. Apparently, changes in shearmodulus, which result approximatelyconsistent with theinference from the deep tow bathymet- fromsample disturbance during coring, and changes in pressureand tem- ric data.At largeO, the effectiveroughness r is muchsmaller than the peratureafter collection are small compared to thenatural variability of wavelengthA, permittinguse of Eckartscattering theory. At small0, surficialsediment shear wave velocity. however,r and ,t. are comparable, violating the Kirchhoffapproximation. Implicationsof theseobservations and othercalculations for modeling reflectivityfrom sediment-free/thinly sedimented basalt, a prevalentfea- ture of the Pacific, will be discussed. 9:05
KK3. Preliminary results from a shallow water bottom interaction 9:50 experimentin the Gulf of Mexico. JamesF. Lynch,George V. Frisk, andSubramaniamD.Rajah (DepartmcntofOceanEngineering, Woods KK6. Selectedexamples of Scholtewave propagation in deepand shallow Hole OceanographicInstitution, Woods Hole, MA 02543) water. Craig A. Fisher and HassanB. All (Naval OceanResearch and DevelopmentActivity, NSTL, MS 39529-5004) In orderto helpassess our presentcapability to predictthe detailsof soundpropagation in shallowwater environments, a cw acoustics experi- The NavalOcean Research and Development Activity recently com- mentwas performed at a sitein theGulf of Mexiconear Corpus Christi, pleteda seriesof threeexperiments to measurevery low-frequency (VLF) TX. Thissite was chosen because it was both relatively simple acoustically seismo-acousticpropagation in bothshallow and deep water marine envi- and previouslywell-studied. Indeed, the sitechosen was the sameas was ronments.A verticalhydrophone array and oceanbottom instruments studiedby Rubanoin an earlierexperiment [L. A. Rubano,J. Acoust. were deployedin all three experiments.The oceanbottom instruments Soc.Am. 67, 1608-1613(1980)]; moreover,a compilationof archival containedboth hydrophone and three-component geophone sensors. In records(seismic data, etc. ) bya NORDA groupgave a secondgeoacous- thesingle shallow water experiment, measurements were also made with tic modelto complementRubano's [J. E. Matthews,P. J. Bucca,and W. buriedthree-component geophones. We comparethe VLF propagation H. Geddes,NORDA Report120, June 1985]. The data were taken at 50 characteristicsin deepand shallowwater seenwith oceanbottom and and140 Hz onradial tracks typically 5 km long,and sampled at leastat watercolumn sensors. The importanceof Scholtewave propagation in thespatial Nyquist frequency (every half-wavelength in range), so that accountingfor bothsignal and ambientnoise energy at nearbottom re- eachdata track was effectively a 5-km synthetic aperture array. The data ceiversis shownfor selectedfrequencies using both measured and mod- werethen Hankel transformed to displaythe normalmode structure of eledresults. Necessary environmental conditions for Scholtewave excita- thewaveguide [G. V. Friskand J. F. Lynch,J. Acoust.Soc. Am. 76, 205- tion are also examined. 216 (1984)], andthe mode structure was in turninverted to yielda geoa- cousticmodel [S. D. Rajah,J. F. Lynch,and G. V. Frisk,J. Acoust.Soc. Am. 82, 998-1017(1987) ], whichwe compareto the NORDA andRu- banomodels. Comparisons of pressurefield predictions, modal structure 10:05 predictions,and geoacoustic models are discussed. [Work supportedby KK7. Gaussian beam method for underwater acoustic reflection from an ONR. ] elasticbottom. I. T. Lu, X. J. Gao, and L. B. Felsen(Department of ElectricalEngineering and ComputerScience, Polytechnic University, Farmingdale,NY 11735)
9:20 The Gaussianbeam method furnishes a potentiallyuseful numerical KK4. Propagationloss modeling of the seabedusing both shearand algorithmfor modelingpropagation in an inhomogeneousocean environ- permeabilityeffects. Mohsen Badiey and Tokuo Yamamoto (Geo- ment. However,the paraxiallyapproximated beams in the heuristically AcousticsLaboratory, University of Miami, Miami, FL 33149) discretizedstack suffer from spectral deficiencies that are problemdepen- dentand make a prioripredictability uncertain. Understanding these defi- A completenormal mode propagation model is used to predicttrans- cienciesmay provide clues as to how to dealwith them.The presenttest missionloss coefficients for two previouslyreported data sets, one from environmentis a line sourceexcited homogeneous ocean with a solid the Nova ScotJanShelf and one from the Mediterranean Sea. This model bottomthat givesrise to two distinctcritical reflectionsassociated with employsboth permeability and shear modulus of theseabed through for- the compressional(P) and vertical($V) waves,respectively. Stacks of mulationsof Blot-Willistheory of acousticwaves in poro-elastic,aniso- narrow beams,which havelocalizing advantages, have previously been tropicmedia. The sediments were modeled using the reported geological foundincapable of accountingfor the criticalangle (head wave) effects andcore data. It wasfound that both shear and hydraulic (permeability) when the relevantwave phenomena involve only P or SV (horizontal propertiesof the sedimentscontribute to the propagationloss. At low shear),respectively [I.T. Lu, L. B. Felsen,and Y. Z. Ruan,Geophys. I. frequencieswaves penetrate deep into the sedimentsand are effected by R. Astron. Soc. 89, 915-932 (1987) ]. It is confirmed here that this situa- thehigher rigidity of thelower sediment layers. Higher frequencies pene- tionpersists in thepresence of P-SVcoupling.As before,very wide parax- trateonly into surficial sediments where the permeability is the dominant ial beamsor narrow beamswith full spectraremove these effects, but influencein the attenuationmechanism. [Work supportedby ONR.] trackingthe latter is computationallymore difficult,while trackingthe former is inconvenient and uncertain for successiveencounters with an environment.Various alternative options are examined to copewith these difficulties.[Work supportedby NOSC.]
9:35
KKS. Low-frequencyscattering from the upper crust. Orest Diachok, 10:20 John Shaffer(Code 5120, Naval ResearchLaboratory, Washington, DC 20375),and Laurel Henderson(SACLANTCEN, LaSpezia,Italy) KK8. The modelingof a shallow-waterwedge using complex rays and steepest-descentintegration. Evan K. Westwood(Applied Research The rms roughnessof sediment-freebasalt at a siteeast of the East Laboratories,The Universityof Texasat Austin, Austin, TX 78713)
S79 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring 1988 115thMeeting: Acoustical Society of America S7!! A newmethod for computingpropagation loss in a shallow-water 11:05 wedgewith a penetrablebottom will bepresented. The methodis based on KKI1. The outgoingsolution of the Klein-Gordon equationand the therepresentation of the total field as a sumof rayfields. Each ray field is expressedasa contourintegral over plane waves in thecomplex 0 plane, effectsof dispersivesediments on pulse propagation in theocean. Michael where0 isincident angle. Saddle points (possibly complex) are found for D. Collins(NORDA NumericalModeling Division, NSTL, MS 39529) eachray, andsteepest descent path integration is requiredwhen the ray The parabolicequation method, which has been used as a frequency strikesthe bottominterface near the critical angle. Lateral wave contribu- domaintool in underwateracoustics for severalyears, is alsoeffective for tionsare also included. This model has been applied to theupslope bench- use in the time domain lB. E. McDonald and W. A. Kuperman, J. markproblems proposed at the 112thMeeting. The resultsare indistin- Acoust.Soc. Am. 81, 1406-1417 (1987)]. In this paper,some recent guishablefrom thosereported by Jensenat the 113thMeeting using progressin thedevelopment ofthe outgoing solution of the Klein-Gordon Evans'two-way coupled mode model (generallyconsidered to be the equationis describedand some calculations are presented.It wasfound mostaccurate). The computation time for 400points on a horizontalline thatvery small amounts of sediment dispersion result in significantdistor- is on the orderof 5 min. Unlike coupledmode and parabolicequation tion of eventhe water-borneportion of signals.This surprisingfinding models,the computation time using this ray modeldecreases asthe fre- shouldhelp resolve a controversyregarding the relationshipbetween at- quencyis raised.[Work supportedby theONR contractN00014-87-K- tenuationand frequency in sediments,which has long been accepted to be 0346.1 linear basedon severalexperiments. In a recentpaper [Zhou et al., J. Acoust.Soc. Am. 82, 2068-2074 (1987) ], a differentrelationship is pro- posedbased on the analysisof arrivaltimes. Despite the factthat attenu- ationand dispersion are very intimately related, dispersion is neglected in the analysis.[ Work supportedby ONR and NORDA. ]
10:35
KK9. Transient Green's tensors for a layered solid half-spacewith different interface conditions.Shuehu Ren, Nelson N. Hsu, and Donald 11:20 G. Eitzen (National Bureauof Standards,Gaithersburg, MD 20899) KK12. Bottomclassification using backscattering at verticalincidence. To understand the influence of the interface conditions on the behav- DarrellR. Jackson(Applied Physics Laboratory, College of Oceanand ior of transientwaves in a structure,the Green'stensor for an isotropic FisherySciences, University of Washington,Seattle, WA 98195) and layer overlayon an isotropichalf-space is solvedusing the "generalized Eric Nesbitt(Boeing Commercial Airplane, M/S 01-41, Seattle,WA ray" expansiontechnique and Willis inversionmethod. This inversion 98124) method has many advantagesover the Cagniard-deHoop inversion A techniquehas been developed for measurement ofbottom geoacous- method.First, the displacementsrather than the potentialsare usedand tic propertiesusing backscattering at vertical incidence. The parameters the derivationsare greatly simplified; second, self-similar sources are as- of a scatteringmodel are determined by a maximum-likelihoodfit to re- sumedand a detaileddiscussion of the structureof a complicatedalgebra- ceivedintensity as a functionof time.The procedureprovides measure- ic function is avoided; and third, the derivativesof the Green's tensor mentsor boundsfor thefollowing sediment bulk acoustic properties; im- aboutspacial coordinates are easily obtained. A FORTRAtacode for com- pedance,absorption coe•cient, and volumescattering strength. In putationof theGreen's tensors was developed. Both the cases of a welded addition,a measureof interfaceroughness can be obtained if thecoherent anda liquidcoupled interface are considered. Computed results show that componentof recievedintensity is negligible. The method has been tested changesof the interfacecondition have a greatinfluence on the behavior usinghigh-frequency data from shallow silty and clayey sites. [Work of transientwaves. For example,when the interface conditions vary from supportedby ONT andNORDA. ] weldedto liquidcoupled, some of the firstfew headwaves nearly vanish while the firstfew regularreflected rays keep unchanged. This phenome- non can be usedto infer the qualityof the interfacebond of materialsin ultrasonic NDE.
11:35
KK13.Time spreadof bottominteracting signals from an abyssalplain environment.Paul J. Vidmar'• (ScienceApplications International Corporation,1710 Goodridge Drive, McLean, VA 22102) 10:50 Measuredsignals from explosive sources are examined to determine somecharacteristics of the time spread introduced by bottominteraction KKI0. Propagation effects of a transition layer at the ocean/sediment in an abyssalplain area. Broadband ( 10-1600Hz) acousticdata are ex- interface. Michael D. Collins, Stanley A. Chin-Bing (NORDA aminedthat sample the transitionfrom sediment penetrating paths that Numerical Modeling Division, NSTL, MS 39529), and Joseph refract within the sediment to those that reflect from the basalt substrate E. Murphy (Departmentof Physics,University of New Orleans,New beneaththe sediment.Different characteristic time spreadsat high and Orleans, LA 70148) low frequencieswere observed. In a high-frequencyband (1500-1600 In thispaper, the resultsof an investigationinto the propagationef- Hz), thetime spread is relativelyshort ( •200 ms), whilein a low-fre- fectsof a transitionlayer in the uppermostportion of the oceansediment quencyband (50-150 Hz) thetime spread is verylong ( >0.5 s). The arepresented. The layerof sedimentdirectly beneath the ocean bottom is shortduration of thehigh-frequency time spread is consistent with reflec- partiallysaturated with waterand is close to thetemperature of the water. tion from layer interfacesin the sediment.The long durationof the time Due to thesesaturation and thermodynamiceffects, the discontinuityin spreadat lowfrequencies isconsistent with interaction with thesubstrate thespeed of sound at the ocean/sediment interface istypically about a half (e.g.,scattering). [Work supported by the NORDA BottomInteraction of a percent.Since the deeper layers are insulated from the ocean and are Projectand the ONR AEAS Program.]'• On leavefrom Applied Re- notsaturated with water,one would expect the sound speed c o deep within searchLaboratories, The Universityof Texasat Austin. thesediment to dependonly on thedepth zbelow the ocean surface to first order. Thus a reasonableformula for sedimentsound speed is c• (z,d) = [ac• (d} - co(d)]e •d-'•/• + co(z), wheream 1, c•ois the ocean soundspeed, d isthe ocean depth, and h isthe thickness of thetransition 11:50 layer.When attenuation is present in thetransition layer, propagation loss increaseswith h sincea thickertransition layer results in longerray path KKI4. The low-frequencyreflection coefficient of a gassysediment. K. lengths within the lossy region. [Work supportedby ONR and E. Gilbertand J. R. Zagar (The NationalCenter for PhysicalAcoustics, NORDA.] Universityof Mississippi,University, MS 38677)
S80 J, Acoust,Soc. Am. Suppl. 1, VoL83, Spring1988 115thMeeting: Acoustical Society of America $80 Narrow-bandpropagation measurements have been made in a shal- frequency.The end points are R = --0.91 +0.02 at 600 Hz and low pondwith a gas-saturatedbottom [ L. E. Murray andK. E. Gilbert,J. R = -- 0.82 + 0.02 at 2000 Hz. Thesevalues for R are similar to those Acoust.Soc. Am. Suppl.I 80, S63 (1986) ]. A normal-modeanalysis of that havebeen obtained in broadbandexperiments [e.g., $on•s et al., $. the data wasused to extractplane-wave reflection coefficients for the Acoust.Soc. Am. 36, 154-157 (1964)]. Physicalmechanisms that may bottom.Between 600 Hz (the cutofffrequency for the firstmode) and controlthe low-frequency reflection coefficient ofa gassysediment will be 2000 Hz, the reflectioncoefficient R variesapproximately linearly with discussed.
THURSDAY MORNING, 19 MAY 1988 418/420 ROOM, 9:30 A.M.
Meeting of AccreditedStandards Committee S12 on Noise
to be held jointly with the
TechnicalAdvisory Group for ISO/TC 43/SC 1 Noise
W. Melnick, Chairman S12 OhioState University, University Hospital Clinic, 456 ClinicDrive, Columbus, Ohio 43210
H. E. von Gierke, Chairman,Technical Advisory Group for ISO/TC 43/SC 1 Director,Biodynamics & BioengineeringDivision, •4F,4MRL/BB U. S. .4JrForce, Wright-Patterson .4FB, Dayton, Ohio 45433
StandardsCommittee S12 on Noise. Working group chairs will report on their progress under the plan for the productionof noisestandards. The interaction with ISOfrC 43/SCi activitieswill be discussed.
THURSDAY MORNING, 19 MAY 1988 EAST BALLROOM A, 10:45 A.M. TO 12:15 P.M.
SessionLL. Structural Acousticsand Vibration IV: Piezoelectric and Damping Materials
Donald Ricketts, Chairman RaytheonCompany, 1847 West Main Road,Portsmouth, Rhode Island 02871-1087
Contributed Papers
10:45 typicalrelaxation behavior in thecomplex elastic coefficients and a peak in thedielectric properties. The transition was studied further by the use LLI. Temperaturedependence of the electromechanicalproperties of of differentialscanning calorimetry in orderto corroboratethe results of ceramicpolymer composite materials for hydrophoneapplications. Kurt the resonancemeasurements. The resultsdemonstrate some difficulties in M. PAttenmyer(Underwater Sound ReferenceDetachment, Naval usingthese particular materials and indicate that other polymeric materi- RescarehLaboratory, P.O. Box 568337,Orlando, FL 32856-8337) alsshould be investigated for usein thesecomposite materials. [Work The real and imaginaryparts of the complexelastic, dielectric, and supportedby ONR. ] piezoelectricproperties of twotypes of commercial piezoelectric compos- ite materialswere studied as a functionof temperaturefrom -- 70 øCto + 50 øCusing an electricalresonance technique described previously [J. 11:00 Acoust.Soc. Am. Suppl.I 82, SI01 (1987)]. Thesematerials consist of ceramicparticles embedded in a polymermaterial and possess improved LL2. Lower symmetryof the elasticand piezoelectrictensors. D. hydrostaticsensitivity as well as mechanical flexibility in comparisonwith F. Nelson (Department of Physics,Woreeater Polytechnic lnatitutc, conventionalpiezoelectric ceramics. Measurements of the temperature Worcester, MA 01609) dependenceof the hydrostaticpiezoelectric coefficients have also been performed.The interrelationship of the properties is complicatedby the A longwavelength lattice dynamical calculation of the elasticand presenceof theglass transition of theelastomeric phase of thecomposite piezoelectrictensors that does not use the adiabatic approximation yields materialwhere sharp changes in the electmechanicaltmosition causes linearcoupling to rotationas well as to strainin dynamicinteractions.
$81 J.Acoust. Soc. Am. Suppl. 1, Vol. 83, Spring 1988 115thMeeting: Acoustical Society of America $81 Thisleads in thegeneral anisotropic (triclinic) case to 45 independent ingthe real and imaginary parts of • aregiven. Experimental results elasticcomponents (rather than 21 as believed for over a century)and to obtainedat roomtemperature are presented for • versuscompresslye 27independent piezoelectric components (rather than 18 ). Thenew con- prestressat twonominal audio frequencies. tributionsto thesetensors result from the inertial effectsof the optic modesand are proportional to the squareof the frequency.Thus they vanishfor staticinteractions and sorestore the traditionalsymmetries in thislimit. The new tensor components ereate an antisymmetric part to the 11:45 stresstensor; it producesa torquethat is balancedby a changein the internalangular momentum density of theoptic modes. The newtensor LLS. Developmentof polyurethaneencapsulants with improved componentswill be measurably large only for frequenciescomparable to resistanceto seawaterexposure. G. M. Stack,J.M. Miller (Underwater opticmode frequencies, which, in mostcrystals, are well beyondthe Sound ReferenceDetachment, Naval ResearchLaboratory, P.O. Box acousticregion. However, near a second-orderphase transition, the fre- 568337, Orlando, FL 32856-8337), and E. Y. Chang (American quencyof a softoptic mode approaches zero. In someeases, such as chlor- CyanamidCo., Bridgewater,NJ 08807) anil,the soft mode has been observed in thehypersonic region as a distinct (underdamped)resonance. In sucheases, the new tensor components can Polyurethaneshave been widely used as underwater acoustic eneapsu- beexpected to bemeasurably large. lantsfor over20 years.In an attemptto producepolyurethane encapsu- lantswith improvedresistance to seawaterexposure, polyurethane sys- temscontaining aliphatic isocyanate groups were investigated. Several commercialprepolymers containing aliphatic isocyanate groups were ob- 11:15 tainedfor evaluation. Initially, low-temperature cure systems were devel- opedto curethese prepolymers. It wasfound that the physical and acous- LL3. Tbermopiezoeleetrieequations of high-frequencyvibrations of ticproperties of theresultant polyurethanes depended significantly on the ecrmroleshells. M. Cengiz D6kmeci (Istanbul TechnicalUniversity, compositionof thecure formulation. The effectof prepolymercomposi- P. K. 9, Taksim, Istanbul 80191, Turkey) tion on the propertiesof the curedurethanes was then investigated. In In view of the author'srecent review article [ ShockVib. Dig. 20 (2), particular,the molecularweight and isocyanatecontent of the pre- 3-20 (1988)], this studyis carriedout for the aim of a systematicand polymerswere varied. Several urethane systems suitable as underwater consistentderivation of thegoverning equations of vibrationsof piezoeer- encapsulantswere developed. These urethanes were subjected to long- mic shells.First, a ¾ariationalprinciple is derivedso as to describeall the termaging in seawater.It wasfound that this exposure to seawater did not three-dimensionalequations of thermopiezoelectricity.Then, the me- significantlyaffect the dynamicmechanical properties of thesepolyure- chanicaldisplacements, electric potential, and temperature increment of thanes.[Work supportedby SonarTransducer Reliability Improvement piezoceramicshell are expandedin powerseries of the thicknesscoordi- Program(STRIP). ] nate.Next, a systemof two-dimensionalequations of successivelyhigher- ordersapproximation is derivedfor vibrationsof thermopiezoelectric shellby useof the variationalprinciple together with the seriesexpan- sions.The systemof equationsgoverns all the typesof coupledvibrations ofpiezoceramicshell [cf. R. D. Mindlin,Int. J.Solids Struet. 10, 625-637 12:00 (1974)]. Also,specia! motions and geometry of piezoceramicshell are indicated.Finally, a theoremof uniquenessis provedfor solutionsof the LL6. Influence of fillers on constrained-layer vibration-damping governingequations of thermopiezoelectricshell. [Work supportedin capabilitiesof chlorobutylelastomers. Rodger N. Capps (Underwater partby theUnited States Army throughits EuropeanResearch Office. ] Sound ReferenceDetachment, Naval ResearchLaboratory, P.O. Box 568337, Orlando, FL 32856-8337)and Linda L. Beumel (Texas ResearchInstitute, 9063 Bee CavesRoad, Austin, TX 78733-6201)
A ehlorobutylelastomer formulation has been developed at thislabo- ratory(USRD) forpotential vibration damping applications. It combines 11:30 the attractive engineeringproperties of low compressionset, good LIA. Evaluationof thecomplex coefficient • of thlckfilm piezoelectric weatherand ozoneresistance, and goodtensile and tear propertieswith desirable viscoelastic behavior for vibration control. An intensive investi- polymers. Donald Ricketts and ThomasHowarth (Raytheon Company,1847 West Main Road,Portsmouth, RI 02871-1087) gationof the effectsof plasticizersand differenttypes and loadingsof fillersupon physical properties and damping capability of the basepoly- A modelfor evaluating•3 ispresented using an impacttesting tech- mer wasconducted. Measurements of thedynamic Young's modulus and nique[J. Acoust.Soc. Am. Suppl.I 67, S32(1980) ]. In theform of a losstangent over an extendedfrequency range were used in conjunction consolidatedhare ring stack, the piezoelectric polymer (PYF 2 ) testspeci- with the Ross-Ungar-Kerwinmodel [Structural Damping, edited by J. menis sandwichedbetween two identicalloading masses. A threadedtie E. Ruzicka (American Societyof MechanicalEngineers, New York, rodpasses through the center of the entire assembly and is fastened at each 1959), Sec.3 ] to predictthe dampingof simplethree-layer constrained endby means of a nut.This arrangement facilitates evaluating • asa structures.The modelpredictions were compared with actualdamping functionof frequencyand compresslye stress. By impact testing this com- measurementsof compositeplates. The observeddifferences between positelongitudinal vibrator, its resonance frequency ( fR ) andmechani- elastomerswith differenttypes of fillersare discussed in termsof mecha- calstorage factor ( Qs•) maybe measured. From these measured quanti- nismsof interactionbetween the fillersand the gum phaseof the elas- tiesand themathematical model of the testresonator, the complexelastic tomer. [Work supportedby Code 1962of David Taylor ResearchCen- compliancecoefficient •3 maybe evaluated. The expressions forevaluat- ter. ]
S82 J. Acoust.Soc. Am. Suppl. 1, VoL83, Spring1988 115thMeeting: Acoustical Society of Amedca S82 THURSDAY AFTERNOON, 19 MAY 1988 GRAND BALLROOM C, 1:00TO 5:15P.M.
SessionMM. SpeechCommunication VII: Cue-Equivalenceand Perceptual Processing
Patricia K. Kuhl, Cochairman Departmentof Speechand Hearing Sciences, University of Washington,Seattle, Washington 98195
SigfridD. Soli, Cochairman BiosciencesLaboratory, 3M Center,St. Paul, Minnesota55144
ContributedPapers
1:O0 Amplitudetransition effects were studied by meansof a formantsyn- thesizer.The signalamplitude control is locatedeither between the vocal MM1. Modelinglisteners' perception of "silent center" syllables. source and the formant filters or between the formant filters and the out- TerranceM. NeareyandJeanE.Andruski (DepartmentofLinguistics, put. In the first configuration,corresponding to the humanvocal tract Universityof Alberta, Edmonton,Alberta T6G 2E7, Canada) situation,at the beginningof the amplitudetransition, an impulsere- Recently,Verbrugge and Rakerd[Language and Speech 29, 39-57 sponseof the formantfilters is obtained.In the secondconfiguration, the (1986)] andStrange and Jenkins [J. Acoust.Soc. Am. Suppl.1 82, S82 formantfilters are alwaysin a stationarystate and, hence,the spectrum (1987)] haveshown evidence that "silentcenter" CVC syllables(i.e., structureof the formantfilter outputs is harmonic.The distortionsintro- syllableswherein only the relatively rapid CV andtransition portions are ducedby theamplitude transitions do not induce more information on the preserved,while the relativelysteady-state midsections are excised)are transferfunction of the formant filters.Clear spectraldifferences were well-perceivedeven when CV andVC portionsfrom different speakers are obtainedwhen comparing the two syntheticsignals, at theironsets. Infor- combined("hybrid syllables"). This studywill presentdata from anala- mationon formantfrequency can be obtainedin the caseof the firstcon- gousexperiments with CanadianEnglish, but includingboth isolated figurationeven if F0 is high. It hasbeen shown that signaldifferences vowelsand/bVb/syllables in aneffort to testthe degree to whichconso- obtainedby the twoconfigurations are specifically processed by the hu- nantalcoartieulation effects (as opposed to vowel-intrinsicproperties) man auditorysystem and inducedisplacements of [i]-[e] boundaries. areresponsible for the highintelligibility of hybridsyllables. The results Moreover,the vowel spectra corresponding to theboundaries do not have obtainedwill be comparedwith predictionsof the patternrecognition a harmoniccomponent in the centerof the firstformant. modelof Neareyand Assmann[J. Acoust.Soc. Am. g0, 1297-1308 (1986)] for CanadianEnglish vowels that exploits vowel-inherent spec- tral change. 1:45
MM4. Interpreting vowel "quality": The dimension of rounding. Leigh Lisker (University of Pennsylvania,Philadelphia, PA 19104and 1:15 Haskins Laboratories, New Haven, CT 06511 )
MM2. Cross-seriesselective adaptation with vowels.Dawn L. Dutton Accordingto one view,speech perception depends crucially on the and JamesR. Sawusch(Department of Psychology,SUNY at Buffalo, articulatoryinterpretation of theacoustic signal, since this would explain Amherst, NY 14260) ß the identificationof the acousticallydifferent initial portionsof two for- Cross-seriesselective adaptation has been found for vowels[J. J. God- mant pseudospeechpatterns such as, e.g., /di/ and /du/: Listeners frey, Percept.Psychophys. 28, 103-111 (1980) ], but the stimulusattri- "know" that theironsets arc producedwith verysimilar tongue gestures. butesrelevant to adaptationwere not clearly inferable. The predictionsof Is thisrelation between perception and articulation versus acoustic signal twomodels of vowelrecognition, the relational model and the octave band typical,or is it insteadrestricted to certainphonetic units? For vowelsit model [D. L. Dutton, J. R. Sawusch,and K. C. Tsoi, J. Acoust.Soc. Am. appearsthat differentcombinations of tongueand lip articulationsyield Suppl.I 81, S18 (1987) ] weretested using selective adaptation. An EH- very similar acousticpatterns and auditory qualities.The 18 "cardinal AE seriesserved as the basecontinuum. Five vowelswere generated to vowels"of Daniel Joneswere presentedto severalgroups of listeners serveas adaptors. The relationshipbetween the testseries and the adapt- askedto decidewhich had beenproduced with lip rounding.Error rates orscould be representedin threeways, depending on the stimulusinfor- varied significantlyover the vowel space,and showeda bias toward re- mationassumed to be relevantfor patternrecognition. The two models portingfront vowdsas unfoundedand backones as rounded,a tendency aboveemploy different frequency-based representations. A third model that presumablyreflects the relativefrequencies across languages of front assumesadaptation is based on thepattern of movementof spectralpeaks unrounded, back rounded, front rounded, and back unfounded vowels. overtime. Adaptors were constructed to besimilar to oneendpoint of the Thus,if speechperception is crucially dependent on an articulatoryinter- test seriesbased on each of thesethree representations.Results will be prctationof an acousticsignal, that interpretationneed not be correct. discussedin termsof what cross-seriesadaptation can tell us aboutthe [ Work supportedby NIH Grant HD-01994 to HaskinsLaboratories. ] representationsthat may be usedduring vowel recognition. [ Work sup- portedby NINCDS. ]
2:o0
MMS. More support for rate-bnseddiscrimination of secondformant 1:30 transitions.M. J. Collins (Kresge Hearing ReaearchLaboratory, LgU Medical Center, New Orleans, LA 70112 and LSU Communication MM3. Effectsof nonstationarycharacteristics on vowel perception. Disorders,New Orleans,LA 70112), J. K. Cullen (KresgeHearing Ren6 Cart6 (Laboratoire de la CommunicationPar16e, ICP Unit6 ResearchLaboratory, LSU Medical Center, New Orleans, LA 70112, Associ6eau CNRS, INPG-ENSERG, 46 AvenueF61ix Viallet, Grenoble, LSU Communication Disorders, New Orleans, LA 70112 and France 38031 ) Departmentof Psychology,University of NewOrleans, New Orleans, LA
S83 J. Acoust.Soc. Am. Suppl. 1, Vol. 83, Spring 1988 115thMeeting: Acoustical Society of America S83 70112), R. J. Porter (Kresge Hearing ResearchLaboratory, LSU Kewley-Portetal. ["Perception ofstatic and dynamic acoustic cues to MedicalCenter, New Orleans,LA 70112and Department of Psychology, placeof articulation in initial stop consonants," J.Acoust. Soc. Am. 73, UniversityofNewOrleans, New Orleans, LA70112),and D. F. Jackson 1779-1783( 1983 ) ] showedthat listeners can reliably identify the place of (Kresge Hearing ReseachLaboratory, LSU Medical Center, New articulationof initialbilabial and alveolarstops in stop-vowelsyllables Orleans, LA 70112) withseverely truncated vowels from just the first 20 msof thesyllable, whichincludes the burstand the beginningof the formanttransitions. Two priorstudies [Besing et al., J. Acoust.Soc. Am. 81, 535 ( 1987); However,they showed that listeners must hear substantially more of the Porteret al., J. Acoust.Soc. Am. Suppl.1 82, S81 (1987) ] investigated formanttransitions, 40 msor more,to identifythe stop'splace of articula- differentialthresholds for onsetfrequency of secondformants (30-ms tionwhen the stop is velar. They speculate that their listeners' need for a transitions)and for transitionduration ( 60- and 120-mstransitions ). The longertransition may reflect a slowerchange in formantfrequencies ob- extentof bothrising and falling (to 1800-Hz)transition was systematical- servedoverall for velarscompared to stopsat otherplaces of articulation. ly variedto yield differingrates-of-change for stimuli.Transition dis- Slowerformant frequency changes for velarsprobably arise from the crimination was examined for formants isolated and in a first formant greaterinertia of thetongue body compared to thetongue tip or lip,as do context.The questionaddressed was whether discrimination varied as a thelonger voice-onset-times (VOTs) observedfor velars versus bilabials functionof the simplestimulus dimensions of frequencyand time or in or alveolars.If a listenerhas other (spectral)cues that a stopis a velar, relationto the morecomplex domain of the rate-of-changeof frequency thenhe will acceptstimuli with longerVOTs as voiced than if theother overtime. Several aspects of the resultssupported a rate-of-changeinter- cuesindicate another place of articulation.Though the longerVOTs are pretation.The present study further tests that interpretation by extending not themselvescues that the stop is a velar,they may enhance that percept observationsto discriminationof onsetfrequency for 60- and 120-mstran- [cf. K. Stevenset al. "Towarda phoneticand phonologicaltheory of sitions.Results for sixlisteners indicate that onset-frequencydiscrimina- redundantfeatures," in lnoarianceand Variabilityin SpeechProcesses, tion of 60-mstransitions is alsobased on rate-of-frequencychange as was editedby J. Perkelland D. Klatt (Erlbaum,Hillsdale, N J, 1986),pp. 426- the casefor 30-mstransitions; discrimination of 120-mstransitions may 449]. Againassuming that other cues to placeare present, slower transi- be, in part, frequencybased. In addition,rate-based discrimination ap- tionsmay enhance the identification of a stopas velar in a similarway, but pearsto be nonlinearin that higherrates-of-change yield smaller differ- couldhave either a negligibleor deleteriouseffect on identifyinga stopas encelimens relative to a referencethan do lowerrates-of-change. [Sup- bilabialor alveolar.In this study,ABX-format discrimination experi- portedby NIH-NINCDS andthe Louisiana Lions Eye Foundation.] mentsare being conducted in whichlisteners hear short ( 100ms) CV syllablesin whichthe second and third formant transitions match natural valuesfor thesyllables [ba, da, ga, bi, di, gi] in directionand magnitude, whiletheir rate of changeis varied in 15-mssteps across a rangeof • 5-90 ms.The stop bursts have been excised from these stimuli so the only cues 2:15 to placeare in theformant transitions. It is expected that listeners' ability MM6. Rate normalization for the/ba-wa/contrast. Rebecca E. Eilers, to discriminateplace of articulationto degradein two directions,with D. KimbroughOller, andRebecca Burns (Departments of Psychology artificiallyshort transitions for velarsand artificially long transitions for and Pediatrics,P.O. Box016820, University of Miami, Miami, FL 33101) nonvelars. A controversialhypothesis in speechperception posits an important rolefor ratecontext in determiningthe boundarylocation between/ba/ and/wa/. In general,Miller and Liberman(1979) suggestthat longer formanttransitions are necessaryto cue/war whenthese syllables con- tainlong rather than short vowels while Shinn et al. ( 1985) suggeststhat boundarylocation is influenced by ratecontext only under special circum- stances.The presentwork reexamines the ratecontext hypothesis using foursynthetic/ha/to/wa/continua differing in vowelduration and four continuavarying in both durationand auxiliarysynthesis parameters. Adult subjectsparticipated in severalstandard and nonstandard percep- 2:45 tion tasksincluding, identification, ABX discrimination,same-different discrimination,and VRISD, a repeatingbackground detection task. Re- MM8. An evaluationof psychophysicaltrading relations and perceptual suitsindicate a small rate effectthat may be largelyaccounted for by the cues. Richard E. Pastoreand Jody Kaplan Layer (Departmentof shortestvowel durations.Examination of individualsubject data across Psychology,SUNY at Binghamton,Binghamton, NY 13901) taskssuggests that the effect may not have important linguistic ramifica- Tradingrelations have been the empiricaland theoreticalfocus of a tions.[Work supported by NIH # HD21534.] numberof researchstudies over the last decade.This paper providesa carefulexamination of the conceptof a cue, the conceptof a trading relation betweencues for a given perceptualcontrast, and published claimsfor empiricaldemonstrations of tradingrelations. While truetrad- ingrelations should be expected for mosttypes of perceptualphenomena, it was discoveredthat many studiesclaiming to find a trading relation 2:30 haveonly demonstratedthe effectsof a singlecue. Furthermore,some aspectsof tradingrelations may be basedupon a failureto adequately MM7. The role of rate of transition in the perception of place of considercritical underlying assumptions about the natureof the contin- articulation.Renee A. E. Zakia and John Kingston(Department of uumalong which the dependent variable is defined. The theoretical impli- Modern Languagesand Linguistics,Morrill Hall, Cornell University, cationsof tradingrelations will bediscussed. [Research supported in part Ithaca, NY 14853-4701) by a NSF grantto the firstauthor. ]
3:00-3:15
Break
S84 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115th Meeting:Acoustical Society of America $84 3:15 aticgroup differences. Several recent findings have favored a distinctpho- neticmode in perceptionof sinewavespeech. Best and Studdeft-Kennedy MM9. lntegrality and trading relations across syllabic and lexical [Proc. 10th Int. Congr.Phon. Sci (1983)] useda labelingtest with a boundaries.Gall R. Tomiak and JamesR. Sawusch(Department of sinewave-analog/ra/-/la/continuum,and found that speech-biased lis- Psychology,Park Hall, State Universityof New York at Buffalo, tenersperformed in a clearlycategorical manner, whereas "music"-biased Amherst, NY 14260) listenersperformed near chance. The presentresearch strengthened that Previousresearch using a speededclassification task with noise-tone finding,by employingAXB identificationwithout overt labels, and AXB analogsof monosyllablesand disyllableshas shown that monosyllabic discrimination. Control tests with a full-formant/ra/-/la/continuum stimulishow integral processing of adjacentphones when perceivedas and sinewaveanalogs of their isolatedF3 trajectorieswere included.The speech,but separableprocessing of adjacentpitch and amplitudeinforma- two groupsperformed identically with eachcontrol continuum. yet they tion whenheard as nonspeech.Furthermore, this integralityeffect was differeddramatically with the sinewavesyllables. In supportof the pho- foundto be morepronounced within syllablesthan acrosssyllable boun- netichypothesis, only the speech-biased listeners systematically classified daries.It wassuggested that thispattern of resultsreflects a speechmode thestimuli into two clearcategories, and showed improved discrimination of processing.The purposeof the presentexperiment was to determine aroundthe catagoryboundary. [Work supportedby NIH grant HD- whethertrading relations can also be attributedto languagespecific cod- 01994 to Haskins Laboratories.] ingprocesses or, to theauditory integration of information spanning sylla- ble (andlexical) boundaries. Analogs of speechstimuli known to produce a tradingrelation [Repp etaL, I. Exp. Psychol.:Hum. Percept.Perform. 4, 621-637 (1978) ] wereconstructed and testedin a speededclassifica- tiontask. The results,and implications of thesefindings, will bediscussed. [Work supportedby NINCDS. ] 4:00 MM 12.Can simultaneous presentation of printinfluence the detection of speechin noise?Ram Frostand Bruno H. Repp(Haskins Laboratories, 3:30 270 Crown Street, New Haven, CT 06511-6695) MM10. Trading relations in intonation. Gerald McRoberts and Whenspoken words are masked by noise having the same amplitude Donald Shankweiler (Haskins Laboratories, 270 Crown Street, New envelope,subjects report they hear the word much more clearly when they Haven, CT 06511 and Department of Psychology, University of seeits printed version at the same time. Using signal detection methodolo- Connecticut, Storrs, CT 06268) gy,it wasinvestigated whether this subjective impression reflects a change in perceptualsensitivity or in bias.Signal-plus-noise andnoise-only trials Tradingrelations inthe'perception ofspeech have been reported both wereaccompanied by matchingprint, nonmatching(but structurally for segmentalaspects of speech[ Best et al., Percept.Psychophys. 29, 191- similar}print, or a neutralvisual stimulus. These results revealed a strong 211 ( 1981) ] andfor intonation,a suprasegmentalaspect of speech[ Stud- biaseffect: The matching visual input apparently made the masking noise deft-Kennedyand Hadding,Lung. Speech 16, 293-313 (1973)]. How- soundmore speechlike, but it did not improvethe detectability of the ever,trading relations in speechproduction have not beenreported. Two speech.(However, reaction times of correctdetections were reliably studiesdemonstrate this phenomenon in the productionof intonation.In shorterin thematching condition, suggesting perhaps a subliminal facili- experiment1, male speakersproduced two sentences("November" and tation.)The biaseffect was much smaller when nonwords were substitut- "I did it") bothas questions and as statements.For questions,a trading ed for thewords. Thus it seemsthat subjects automatically detect corre- relationbetween the peakFO andthe amount of FO riseon thefinal sylla- spondencesbetween speech amplitude envelopes and printed stimuli, and ble wasfound: As peakF0 increased,the amountof F0 riseon the final theydo this more efficiently when the printed stimuli are real words. This syllabledecreased. This result is consistent with the findings in perception supportsthe hypothesis, much discussed in the reading literature, that and suggestsa link betweenthe perceptionand productionof intonation. printedwords are immediately translated into an internal representation In experiment2, maleactors produced questions and statements using the havingspeechlike characteristics. [ Work supported by NICHD. ] sametwo sentences.Peak F0 wasvaried in two ways:linguistically (em- phatic/nonemphaticstress), or affectively(simulated happy and sad emotions).Utterance types in eachcondition were verified by listeners. The existenceof a tradingrelation between peak F0 andF0 finalrise for questionswas confirmed for the emphaticstress condition, but notfor the affectirecondition. This dissociationconfirms indications from percep- 4:15 tual studiesof a linguisticorigin for theeffect. These results demonstrate a perception-productionlink for linguisticallyrelevant uses of intonation. MMI3. The interactionof v'sualplace and auditory voicing information [ Work supportedby NIH •NS-24655 and # HD- 1994.] duringthe perception of speech.Kerry P. Greenand Patricia K. Kuhl (Departmentof Speechand Hearing Sciences, University of Washington, Seattle,WA 98195) Previousstudies using speeded classification paradigms have been 3.-45 usedto testwhether the information that underlies the perception of place MM11. Perceptualeffects of attentionto phoneticversus "musical" andvoicing is processedin an integral as opposed to a separatefashion. Resultsof auditoryonly experiments show that reaction times (RTs) for propertiesofalnewavespeech.CatherineT. Best(Haskins Laboratories, 2?0Crown Street, New Haven,CT 065] l andDepartment of Psychology, placeclassification in a conditionin whichstimuli vary along both place andvoicing (the orthogonal condition) are longer than RTs in a control WesleyanUniversity, Middletown, c'r 0645?), Michael Studdcrt- conditionin whichstimuli vary only in place.These data are takenin Kennedy(Haskins Laboratories, 2?0 Crown Strcet,New Haven, CT 0651], Departmentof Linguistics,Yale University,New Haven,CT supportof the idea that the information underlying place and voicing axe integrallyprocessed. This study was concerned with whether or not place 06520,and Department of Psychology,University of Connecticut,Storrs, classificationwould demonstrale the samepattern of resultswhen the CT 06268), and Sharon Manuel (MIT Speech Laboratory, MIT, placefeature was determined byboth auditory and visual (the face of the Cambridge,NLA 02]39) talker)information. To dothis, the auditory tokens/ibi/and/ipi/were A continuingdebate in speechperception research has been whether pairedwith a videodisplay of a talkersaying/ibi/or/igi/, and presented phenomenasuch as categorical perception and trading relations are ex- to subjectsfor speeded classification of place or voicing.Because of the plainedby basicpsychoacoustie mechanisms or ratherby a speciali•-ed "MeGurk effect,"these stimuli were perceived as/ibi, ipi, idi, or iti/. The phoneticmode of perception.Sinewave speech analogs have provided an resultsshowed that RTs for classifyingphonetic features in the orthogo- opportunityto testthese alternatives, since the same stimuli can be heard nalcondition were signficantly longer than in thecontrol condition. These as eitherdistorted speech or nonspeechsounds by differentgroups of resultswere obtained for bothplace and voicing judgments. Thus audi- listeners.The psychoacoustic hypothesis would predict similar between- tory-visualresults appear to mirrorthose obtained in previousstudies groupperformance, while the phonetic hypothesis would predict system- usingonly auditory stimuli. [Work supportedby NIH.]
885 J. Acoust.Soc. Am. Suppl. 1, VoL83, Spring 1988 115thMeeting: Acoustical Socioly of Amodca S85 4:30 gibility.Furthermore, it wasshown that these results could be fairly well predictedby articulationtheory. The presentstudy extends this work by MMI4, Factors affecting the integrntinn of auditory and visual examingthe theory'sability to predictauditory-visual intelligibility for information in speech:The level effect. Patricia K. Kuhl, Kerry filteredbands of speechin noiseas well as for variouscombinations of P. Green (Departmentof Speechand HearingSciences), and Andrew nonadjacent1/3 octavebands in quiet.A comparisonof predictedand N. Meltzoff (Departmentof Psychology,University of Washington, obtainedintelligibility results will be discussedfor eachfiltered band as a Seattle, WA 98195) functionof speech-to-noiseratio. In addition,the assumptionthat the The "McGurk effect"is a phenomenonin whichan illusorysyllable is auditoryAI obtainedfor eachseparate band can be addedto predictthe perceivedwhen discrepant auditory and visual information are combined. auditoryand auditory-visualAI when the bandsare combinedwill be For example,an illusory/dayis perceived when an auditory syllable foa/ tested.[Work supportedby NIH.] is pairedwith a videodisplay of a talkersaying/gaY. In a seriesof studies reportedhere, it wasfound that thelevel of theauditory signal substantial- ly affectsthe illusion.Moreover, this effectof auditorylevel works in a directionthat is counterintuitive.One might have postulated that as the levelof the auditorysignal is increased,making the auditorysignal more pronounced,the illusionwould decrease. These results indicate just the opposite.As the level of the auditorysignal goes up, rather than it 5:00 outweighingthe visualsignal and thereby decreasing the illusion,the ef- fectis to increasethe numberof illusoryresponses. Paired were the same MMI6. Choosingan appropriatetechnique for measuringthe benefitof auditorysyllable/bay at threelevels, soft (45 dB SPL), moderate(58 dB lipreading to speech perception in noise. Quentin Summerfield, SPL), and loud (66 dB SPL), with a videodisplay of/gay. Significant Alison MacLeod, Diana Field, and John Foster (MRC Institute of increasesin illusory/daY responsesoccurred as the auditorysignal in- HearingResearch, University Park, NottinghamNG7 2RD, England) creasedfrom soft to moderateand again from moderate to loud.The work thusshows that auditory level is an importantdeterminant in the percep- The reliabilityof two proceduresfor measuringthe benefitprovided tion of the illusion.The paradoxof the situationis that as the auditory by lipreadingand speech perception in noisewere compared. Benefit was signalincreases in strength,the resultingpercept becomes less veridieal measuredas the differencebetween two signal-to-noiseratios (SNRs) ratherthan more so. [Work supportedby NIH.] givinga criterionlevel of intelligibilityagainst a fixedlevel of white noise whenspeech was presented: (i) in audio-aloneand (ii) in audio-visual conditions.In the first technique,sentences were presented using an up- downrule to convergeon the SNR giving50% intelligibility.In the sec- ondtechnique, continuous speech was presented and subjects adjusted the 4:45 SNR until theycould "just understand every word that wasspoken." For MMIS. Further studieson the auditory-visualarticulation index. Ken the sameduration of testing,the sentence-basedprocedure yielded an W. Grant (ResearchLaboratory of Electronics,MIT, Cambridge,MA estimateof benefitthat differedmore reliablyfrom zero and that had a 02139) greatertest-retest reliability. This estimatealso correlated more highly with an independentestimate of a subject'sability to lipreadsentences Previouswork on the auditory-visualarticulation index [Ken W. with visionalone, probably because the sentence-based procedure places a Grant,J. Aeoust.Soc. Am. Suppl.I 79, S4 (1987) ] showedthat filtered higherpremium on the useof visualcues. Overall, the sentence-based bandsof speechwith differentcenter frequencies but roughlythe same procedureis the bettermeasurement tool, althoughthe self-adjustment auditoryintelligibility in quietalso have the sameauditory-visual intelli- proceduremay have some value in rehabilitation.
THURSDAY AFTERNOON, 19 MAY 1988 WEST BALLROOM B, 1:20 TO 2:55 P.M.
Session NN. Underwater Acoustics VI: Underwater Noise
David J. Thomson, Chairman DefenceResearch Establishment Pacific, FMO Victoria,British Columbia V8W 2Y2, Canada
Chairman's Introduction--l:20
ContributedPapers
1:25 The sourceswere located in the deep-waterbasin and overthe shallow- watershelf. Propagation for a summerSargasso sound-speed profile was NNI. Topographicnoise stFipping at obliqueangles over continental consideredfor sourcesat variousranges from the coastlineand locatedin slopes. Donald R. Del Balzo, Deborah L. Head, and Mona the range-depthplane normal to the coastline.The presenceof a noise J. Authement(Naval Researchand Development Activity, NSTL, MS notch,a regionof relativelylow noise,at shallowdepths over the slope, 39529) wasdiscovered. In this presentation,the analysiswill be expandedto includeseasonal effects (summer and winter sound-speed profiles), fre- In a previouspresentation [ D. R. Del Balzo,M. J. Authement,and C. quencydependence (30 and 100Hz), andsources located at obliquean- T. Mire, J. Aeoust.Soc. Am. Suppl.I 78, S3 (1985) ], the effectof a 2.8- glesout of the range-depthplane normal to the coastline.The effectof degslope on the noisefield generatedby 100-Hz sourceswas discussed. theseparameters on the noisefield and noisenotch will be discussed.
S86 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115thMeeting: Acoustical Society of America $86 1:40 for serialcorrelation. Results suggest that the burstevent point process may be modeledas a piecewiseconstant nonhomogeneous Poisson pro- NN2. Temporaland spatialcorrelations for Iow-freqeuncyambient noise cess.This model is acceptedat > 5% confidencefor all testsapplied. in a diffuseice edge.N. C. Makris (Departmentof OceanEngineering, Thougha limited data setis utilized,the analysismethodology developed MIT, Cambridge,MA 02139) is applicableto largersets of data. Spectralcontent of tonal components Low-frequencyambient noise data acquiredduring Mizex 84 have are alsoexamined using Thompson's multiple window technique, and a beenanalyzed for temporaland spatialvariations. Several concurrent harmonicrelationship is found to exist. time seriesof rms pressurein the 25- to 50-Hz bandsampled at 30-s intervalsand of a few hoursin durationwere constructedfrom hydro- phonedata taken at variousseparations within a 10-km range.During noiseepisodes, a time scaleon the order of 10 min hasbeen obtained via autocorrelations.Also, variationsin noiselevel of up to 7 dB between sensorsas close as 1/2 kin, togetherwith crosscorrelationswhich are high at separationsof 5 km but significantlysmaller at half that distance,sug- 2:25 gesta spatiallybanded or striatednoise generation process. [Work sup- portedby the Officeof Naval Research.] NNS. Distribution of surface-generatednoise in a three-dimensional environment.John S. Perkins, W. A. Kuperman, F. Ingenito (Naval Research Laboratory, Washington, DC 20375), and Michael P. Porter(SACLANT UnderseaCentre, 1-19026, La Spezia,Italy) Modelingthe distribution of noisegenerated at theocean surface over a wideocean area with a spatiallyvariable source level distribution in a three-dimensional(3D) environmentrequires the capability of efficiently computingthe acoustic field from all pointson thesurface to all pointsin the ocean.A previouslydeveloped noise theory is modified[W. A. Ku- 1:55 permanand F. IngenitoJ. Acoust.Soc. Am. 67, 1988-1996(1985)] and NN3. Analysis of marginal ice zone noise events. Chi-Fang Chen is combinedwith an approximatenumerical technique for computing (Departmentof OceanEngineering, MIT, Cambridge,MA 02139) transmissionloss in a 3-D environmentbased on adiabatic mode theory [M.P. Porter,W. A. Kuperman,F. Ingenito,and A. Piacsek,J. Acoust. Underwater ambient noise data from MIZEX 84 wcrc examined and Soc.Am. Suppl.1 81, S9 (1987}]. The waveequation solution of the transient noise events were detected and studied. Event locations were acousticfield (or its cross-spectraldensity) from any pointto any point determinedand their peak valuesused to calculatethe sourcestrength (or to anypair of points)reduces to a "spreadsheet"type manipulation of basedon a simplevertical dipole model. The resultsindicate that the precomputedmodal solutions.This 3-D range-dependentnumerical sourcestrength isabout the same order of magnitudeasthe strength of schemeis shownto reduceto analyticexpressions for the distributionof sourcesat correspondingfrequency in the centralArctic. Event signa- noisein a horizontallystratified ocean. tureswere classified, but, in contrastto the centralArctic case,are consid- erablylonger than their scaled counterparts. [ Work supportedby Office of Naval Research.]
2:40
NN6. An analysisof low-frequencywind-generated ambient noise source
2:10 levelsin theocean. W. M. Carey,D. J. Kewley,'• D.G. Browning,and W. A. VonWinkle (New London Laboratory, Naval Underwater NN4. Statistical characterization of Arctic under-ice acoustic bursts and SystemsCenter, New London,CT 06320) tonals.Samuel M. Nagle, Clinton T. Siedenburg(Honeywell, MSD, Historically,the greatest amount of publishedambient noise data has 6500Harbour Heights Parkway, Everett, WA 98204-8899),and James beenobtained in NorthernHemisphere Oceans. However, principally due A. Ritcey(Department of ElectricalEngineering, FT-10, Universityof to shipnoise contamination and lack of directionality,only a limitedpor- Washington,Seattle, W A 98195) tion canbe usedfor determininglow-frequency and wind-generatedam- A 10-min time series of FRAM II Arctic under-ice noise data is ana- bient noise source levels. Some data have been extracted and combined lyzed,extending the results of an earlierstudy [J. G. Veitch,A. R. Wilks, with theSouthern Hemisphere database [A. S.Burgess and D. J. Kewley, andS.C. Schwartz,"Characterization of Arctic underseanoise," pp. 45, J. Acoust.Soc. Am. 73, 201-210 (1983)]. Theseresulting wind-genera- Dep.Stat., Princeton University (June 1983), AD-AI30 397/3 (No. 22, ted ambientnoise source levels are difficult to comparedue to differences 1983)]. Specificanalysis is conductedon the impulsivenoise bursts and in the sourcelevel definition and descriptions by severalinvestigators. A tonals that are characteristic of underwater noise when an ice cover is standardset of sourcelevels is proposed based on a monopolebeneath the present.Point processmodeling of the timesof arrival of the impulse pressurerelease surface and an equivalentdipole at the surface.The am- burstsis performed by testing the empirical distribution of theburst event bientnoise source level data reported in the literatureare shown to be in interarrivaltimes against an exponential parent distribution, characteris- generalagreement. a•Present address: Weapons Systems Research Labo- tic ofa Poissonprocess. Goodness-of-fit tests are invoked, as well asa test ratory, DSTO, GPO Box 2151, Adelaide,S.A., 5001,Australia.
S87 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S87 THURSDAY AFTERNOON, 19 MAY 1988 WEST BALLROOM A, 1:25 TO 5:20 P.M.
SessionOO. PhysicalAcoustics VI and Bioresponseto Vibration III: ExtracorporealShock Wave Lithotripsy--Biological Aspects
David T. Blackstock, Chairman AppliedResearch Laboratory, University of Texas,P.O. Box 8029,/lustin, Texas78713-8029
Chairman's Introduction--l:25
Invited Papers
1:30
OOI.Cavitationinlithotripsy.Edwin L. Carstensen(DcpartmentsofElectricalEnginecring, Biophysicsand the RochesterCenter for BiomedicalUltrasound, University of Rochester,Rochester, NY 14627) The pressureamplitudes of theshock waves used in lithotripsyfar exceedthresholds for transientcavita- tion.Where appropriate nuclei exist within the body, it ishighly probable that bubbles will formand collapse violently,giving risc to potentiallyundesirable side effects of the treatment.Evidence from basicand clinical studiesin support of this conclusion isaccumulating rapidly. Drosophila larvae have served as useful biological modelsfor studiesof cavitation-relatedphenomena because of thehighly reproducible microbubbles within theirrespiratory system. Using approximately 100 000, l-ps pulsesfrom a 2-MHz piezoelectricsource, the thresholdfor killingof larvaeoccurs at a pressureamplitude of -0.7 MPa. With an electrohydraulic(unfo- cusedspark source) lithotriptcr, three positive spikes of • 1 MPa aresufficient to kill roughlyone-half of the exposedlarvae. This is two orders of magnitude less than the largest reported pressures for clinical lithotripters.
2:05
002. Biologicaleffects of shockwaves. M. Delius(Institute for SurgicalResearch, University of Munich, Munich, FederalRepublic of Germany), H. Eizenhoefer,R. Denk (Dornier MedizinteehnikGmbH, Gemering,Federal Republic of Germany),H. Liebich(Veterinary Anatomy, University of Munich, Munich, FederalRepublic of Germany),and W. Brendel(Institute for SurgicalResearch, University of Munich, Munich, FederalRepublic of Germany) Theeffect of shockwaves on normalcanine kidneys was examined in twogroups of dogswhose right kidneyswere exposed to 3000shock waves in a DornierHM II lithotripter.The groups differed only in therate of shockwave administration, which was 100 and I persecond, respectively. At autopsy, more hemorrhages andhemolysis occurred if shock waves were administered at a rateof 100per second. In anotherexperiment, extracorporealshock waves were slowly administered to piglet liver. They induced two types of changesas detectedby conventionalreal-time ultrasound: Transient changes consisted of burstsof bubblesin liverveins thatwere flushed away with the blood flow. Permanent changes, which appeared later after several hundred shockwaves, consisted of a brightarea of increasedair contentin thehigh-pressure field of thewave. This is evidencethat shockwaves generate cavitation in vivo.Supported by furtherexperiments, a model of shock waveaction is proposed:The negativepressure part of the shockwave generates microbubbles which, if encounteredby thepositive pressure part of thefollowing shock wave, generate the shock wave effects. [ Work supported,in part, by Kurt Kfirber Stiftung.]
2:40
003. MateriMs fragmentationby shockwaves. B. Finlayson(Division of Urology, Box J-247 JHMHC, Universityof Florida, Gainesville, FL 32610),Mohamed Nasr ( 11Medan EI-Tahreer No. 22,Cairo, Egypt), and J. Paul Whelan (29 HeritageDrive, # 19, StoneyCreek, Ontario LSG 4T4, Canada) Althoughextracorporeal shock wave lithotripsy (ESWL) is an importanttechnique, the detailsof howit works•how shockwaves interact with stonesto causedisruption--and how to optimizeits effectare not known with certaintyby practitionersof the art. Three position-dependentmechanisms of disruptionwere observed.Position dependence was probably due to focusingof the shock wave by an ellipsoidal reflector. Near the focus,the frontsurfaces of the testobjects were roughed (spalated), probably by collapseof cavitation bubblescaused by passageof the shockwave. Also, midspecimen cleavage was seen, presumably caused by back wall reflectiontensile stress. Distal to the focusback wall, spalationoccurred, possibly due to tension causedby the expandingshock wave. Sieving fraction observations made of the effectof shockwaves on "z brick" showthat fragmentationhas an approximatefirst-order dependence on the numberof shocksand a second-orderdependence on the voltage at whichthe shocks were generated. The multiphase character of thez brickwas distinctly revealed in plotsof percentfragmentation versus shock number; z brickis not a satisfactory test material, and there is a need for a well-characterized set of test materiMs.
S88 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S88 3:15
004. Extracorporealshock wave lithotripsy of g•llstones:Objectives, current limitations, and preliminary in vitroand in vivoobservations. $. L. Thistleand B. T. Petersen(Division of Gastroenterology,Mayo Clinic, Rochester,MN 55905)
Surgicalremoval ofgallstones is performed about 500 000 timesper year in theUnited States. Although the usualcost ranges from $5000 to morethan $10 000 per patient,the operativemortality is lessthan 1% for otherwisehealthy persons. Nonsurgical dissolution of gallstonesis oftenpossible, but manygallstones contain calciumcompounds that slowor preventdissolution using the directcontent cholesterol solvent, methyl tert- butylether (MTBE). Extracorporealshock wave lithotripsy (ESL) haspotential clinical utility for facilitating dissolutionand possiblyallowing spontaneous or inducedpassage of smallfragments via the bile ductsand intestines.Obstacles to beovercome include predictable fragmentation of multiplestones up to 3 cm in diame- ter sothat all particlesare small enough to berapidly dissolved or passsafely through the bile ducts ( < 3 mm). This mustbe achievedusing biologically tolerable shock wave dosages without requiring general anesthesia withinone or two treatmentsessions. Acute and chronic safety, patient acceptance, and cost must be competi- tivewith surgicaltreatment. Initial experiencesuggests that only solitary noncalcified stones ( 10% of patients with stones)have a highprobability of successusing ESL plusoral bileacid dissolution therapy. Improving ESL technologyand directcontact fragment dissolution using MTBE shouldincrease the clinicalutility of shockwave lithotripsy of gallstones.
ContributedPapers
3:50 Humanapplication of single-cyclefocused shock waves requires an intensestudy of the interactionprocesses associated with shockwave 005. Cavitationin flowingmedia by lithotripter shockwaves in vitroand propagationin biological media. In thisconnection, shock wave-induced in vivo.Douglas L. Miller (BattelleNorthwest, P.O. Box999, Richland, cavitationprocesses onthe scale of a singlebiological cell are of funda- WA 99352), M. Delius (Institutefor SurgicalResearch, University of mentalimportance. On thebasis of a modelanalysis including different Munich,Munich, Federal Republic of Germany),A. R. Williams propagationmedia for thebubble-shock wave interaction and various (Department of Medical Biophysics,University of Manchester, shockwave profiles, possible hazards are discussed and compared with Manchester, United Kingdom), and W. Schwarze (Dornier- experimentaland clinical results. Medizintechnik,Munich, FederalRepublic of Germany) Cavitationproduced by lithotriptershock waves was characterized in vitroin waterand blood, and in vivoin aorticblood by meansof a resonant bubbledetector. The 1.6-MHz detectorcan detectand count4 _+ l-/.zm- diambubbles flowing through it by receivingtheir second harmonic emis- sionsat 3.2MHz. Spark-gaplithotripters were used to exposethe flowing liquidupstream of thedetector at the shockwave focus. This systemwas readilyable to detectbubbles resulting from shockwave-induced cavita- tion in bothwater and bloodflowing through plastic tubes in vitro,and 4:30 evenin bloodpumped by theheart through a plasticarteriovenous shunt. Multiple(up to severalhundred) bubble counts were obtained for each 007. Determinationof designparameters for an ultrasonickidney stone shockwave. However, for 200-400shock wave exposures of eachof two disintegrator.R. Agarwal and V. R. Singh (Department of dogs,this systemwas unable to detectevidence of shockwave-induced Instrumentation and Biomedical Ultrasonics, National Physical cavitationalactivity occurring within the intactvascular system in vivo. Laboratory,New Delhi, 110012,India) [Work supportedby DornicrMedizintechnik and by NIH Grant CA 42947.] Acoustictechniques are now increasingly used for theremoval of kid- neystones without surgery. In the past,generally, acoustic shock wave techniquesand lithotripsy have been used. However, focused ultrasound 4:10 maybe usedto disintegratesuch stones, without any effect on the sur- roundingtissues. Ultrasonic charcteristics of kidney stones are studied to 006. Cavitation processesinduced by weak shock wavespropagating in determinefurther the designparameters such as modulusof elasticity, tissue. H. Koch, M. Griinewald, and H. Herineking (Domlet acousticimpedance, frequency of vibrations,etc., to developa kidney Medizintechnik GmbH, Industricstr. 15, 8034 Gcrmering, West stonedisintegrator. The valueof ultrasonicvelocity in kidneystone in Germany) vitrois foundto be almostdouble that of water or softbiological tissues.
4:50-5:20 Bull Session
S89 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S89 THURSDAY AFTERNOON, 19 MAY 1988 GRAND BALLROOM A, 1:30 TO 4:50 P.M.
SessionPP. Musical AcousticsIV: Arthur H. BenadeMemorial SessionII: Horns, Strings, and Harmony
ThomasD. Rossing,Chairman Departmentof Physics,Northern Illinois University,DeKalb, Illinois 60115
Chulrman's Introduction--l:30
Invited Papers
1:35
PP1.Reflections on the development ofa physicist.R. E. Chrien(Brookhaven National Laboratory, Physics Department,Upton, NY 11973) We arehere to honorthe contributions of Arthur Benadeto understandingthe physics of musicalinstru- ments.These unique contributions were brought to fruitionby a setof fortunatecoincidences that resultedin a shiftfrom Benade's original career orientation in nuclearphysics to oneof musicalacoustics. As Benade'sfirst student,from 1952 to 1957,I wasin a positionto viewth•s change at firsthand, and perhaps to understandhow it occurredbetter than most others. The conting together of Arthur'sdeep interest in pedagogy,his broad view ofthe unity of physics, and the cooducive atmosphere present in the department at Case•withits historical ties to acoustics•wereresponsible for initiatinghis career in the acousticsof musicalinstruments. Lessons for us all maybe drawn from his ability to combinehis vocation and his avocation in a harmoniousand productive manner.I wouldlike to sharewith you my remembrancesof this formative time in hisprofessional life.
2:05
PP2. A simplifiedmodel for a beatingreed--Preliminary results. W. E. Worman(Department of Physics, MoorheadState University, Moorhead, MN 56560)and A. H. Benade(Physics Department, Case Western ReserveUniversity, Cleveland, OH 44106) Thispaper presents early results of a modelfor thebehavior of a musicalwind instrument with a beating reedin thecase where the air flow is completely blocked for a finitepart of each period. The parameters used are appropriatefor a clarinet.The method is as follows: (a) Assumea waveformfor the transarnndal pressure, Pi, (pressuredifference across the reed);(b) calculatethe reedposition assuming that the reedis a simple harmonicoscillator driven by p•,; (c) usethe pressure and reed position to calculatethe flow into the mouth- piece;(d) usethe flow and input impedance of theair column to calculatethe transarundal pressure, Po.t; and (e) adjustthe assumedPi, sothat the calculatedPou• agrees with it. The resultsobtained so far allowsome generalconclusions. The pressurewaveform is "squarish,"almost symmetrical, and varies between about zero andtwice the blowingpressure. The behavioris independentof systemparameters over wide limits and is controlledby energyexchanges that occurwhen the transarundalpressure is nearits minimum.
2:35
PP3. RecoUecfionsof studiesin the laboratoryof A. H. Benade:1975-1978. Stephen C. Thompson (Acousticsand SensorSystems, Ocean Systems Division, Gould Inc., 18901Euclid Avenue, Cleveland, OH 44117)
From 1975to 1978th• authorhad the honorofatudying acou•tica aa a graduate•tudcnt in thc laboratoryof ProfessorA. H. Benade.By thistime, much of the fundamentalwork in understandingthe physicsof wind instrument air columns,tone holes, and the nonlinear interaction betweenthe air column and its excitation mechanismhad been developed. Fundamentals of MusicalAcoustics had just beensent to the publisher. Questionsconcerning some of the more subtle aspects ofwind instrument behavior were beginning tobe asked: reedresonance effects, spectral transformations from the instrumentto the room, mutual interactionof the radiationfrom adjacenttone holes,and the modificationof air columnresonances by the player'sairway impedance.And all of thenew physical understanding was routinely being converted to themusical instrument craftsman'sintuitive methodsfor improvingthe playingbehavior of instruments.For Art, the theoretical understandingof physicalphenomena and the practical application of thesenew ideas in the"real world" of the musicianheld equal statns. This constantinterplay of theory,experiment, and practical application created a chain reactionof ideaswhich is still underway.This paper will recountthe author'srecollections of the atmosphereand developmentsof that time period.
S90 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S90 3:O$
PP4. Woodwinddesign algorithms to achievedesired tuning. Douglas H. Kecfe(Systematic Musicology Program,School of Music,DN-10, Universityof Washington,Seattle, WA 98195) Techniquesexist to analyzewoodwinds of knowngeometry. The inverse problem is a techniqueby synthe- siswhich determines tone hole placements to approximateideal tuning for tonesthe instrument is intended to produce.Two algorithmsusing a relaxationtechnique have as inputs:mouthpiece equivalent length, bore shape,tone hole dimensions, and modenumber. Outputs include the tonehole positions,deviation from harmonicalignment of modalresonances, and open hole lattice cutoff frequencies. Bad designs are detected, forcingthe designer to modifythe bore or tonehole geometry. The first algorithm, due to Benade,uses a model basedupon regular, semiinfinite tone hole lattices, and he employed it in designingthe NX clarinetand other woodwinds.The newalgorithm incorporates damping, a generaltone hole model, lattice irregularities and finitebell termination.The algorithmallows calculation of input impedanceand reflectionfunctions for subsequenttime domain simulations. The playing behavior of flutesconstructed using these algorithms corre- lateswell with predictions.Comparisons of the algorithmsand their sensitivityto variationsin tonehole impedancemodels will be discussed.
3:35
PP$. Frommusician to listenervia first principlesand careful observation. Peter L. Hoekje(Department of Medicine,Case Western Reserve University, Cleveland, OH 44106) Manyof theelements comprising the acoustics of musical wind instruments have been known since the time of Lord Rayleighand Helmholtz. Yet, onlyin thelast few decades have the contributions of acousticiansbeen ableto weighsignificantly with thoseof artisans(whose names like Boehm,Conn, Hottetcrre,or Sax have becomefamiliar to us throughthe instrument-makingdynasties founded on lheir acousticinsights) in the creationof fineinstruments. Art Benadewas a leaderin theserecent successes. Part of thelegacy he leaves us is a methodologyfor design(or redesign)of wind instruments.It describesthe adjustmentof toneholes, bore shapes,and reeds in orderto improvemode cooperation and decrease energy loss, verified by a seriesof simple playingexperiments. By his unique insight, theoretical results arising from fundamental physical principles are identifiedwith commonattributes of "good"instruments. By similarstudy, he improvedthe understanding of the perceptionof soundin rooms.The life work of Art Benadewas the isolationof factorsof importanceto musiciansby soundscientific practice, inspired observation, and carefulexperimentation.
4.'05
PP6. A. H. Benadeand the perceptionof music.Ian M. Linderaid (Institute of Electroacoustics,Technical Universityof Munich, Arcistrasse21, D-8000 Milnehen2, WestGermany) Benade'sfascination with thegreat skill by which the auditory system understands the complicated signals suppliedto it grewfrom his primary interest in thephysics of musicalinstruments. Often, solutions to problems of physicswould force questions of perception.The link betweenphysics and perception was natural for him, roundingout his knowledgeof musicalinstruments as soundsources and the concerthall as a transmission systemwith seriousinquiry regarding the auditorysystem as the sound'sreceiver. Benade's work in musical perceptionwas guided by hisdevotion to "realworld" acoustics and reflected his eommitment to approaching his sciencefrom the viewpointof thosemost affected by music,the musiciansand experiencedlisteners. The author,who was Benade's last degree-earning doctoral student, will discussthe work in musicperception: from the coffeeroom queries, through the formal research,and, finally,on to the continuingefforts to answerthe questionsthat Benareleft, so that we too may havesome fun. [Work supportedby the ASA's F. V. Hunt PostdoctoralFellowship.]
ContributedPaper
4:35 attitudetoward science,profound knowledge of physicsand acoustics, PP7. Somememories of A. H. Benade.Zili Li (Departmentof Physics, uniqueability doing both theory and experiment,and the combinationof CaseWestern Reserve University, Cleveland, OH 44106) humorouslanguages and pleasantpersonality gave me an unforgettable impression,especially for a studentfrom China. I'll recall someof my Dr. Benadowas my thesisadvisor after I gotmy candidacyfor Ph.D. experiencesworking with him,starting with myfirst impression of him to in May 1985and we have workedtogether since then. Art's rigorous the last dayswe spenttogether in our laboratory.
S91 J. Acoust.Soc. Am. Suppl. 1, VoL83, Spring 1988 115thMeeting: Acoustical Society of America S91 THURSDAY AFTERNOON, 19 MAY 1988 ASPEN ROOM, 1:30 TO 3:35 P.M.
SessionQQ. Noise V: Noise Measurementand Propagation
GovindappaKrishnappa, Chairman EngineeringLab. M7, NationalResearch Council of Canada,Montreal Road, Ottawa,Ontario K1A OR6, Canada
Chairman's Introduction--l:30
ContributedPapers
1:35 Burgess[J. Acoust.Soc. Am. 70, 715-726 ( 1981) ] simulatedan ac- tivenoise reduction system in a duct.This system utilized a versionof the QQI. Scattering/diffractioneffects of a sound intensity probe filtered-xLMS algorithmto obtain noisereduction over a wide band- incorporatingdirect particlevelocity measurements. G. Krishnappa width.A similarsystem has been implemented using the TMS-32020 sig- (Division of MechanicalEngineering, National ResearchCouncil, nalprocessing chip on a PC expansionboard. In thefiltered-x LMS algo- Ottawa, Ontario K 1A 0R6, Canada) rithm, the referencesignal must be filtered by the transfer function Experimentalstudies were carried out to examinethe scattering/dif- betweenthe transversalfilter output and the errorsignal input in orderto fractioneffects of a soundintensity probe that determines particle velocity maintainstability of the LMS weightupdate. In our system,this transfer directlyby usingtwo pairsof ultrasonictransmitters and receivers.The functionis calculated from the measured response to a maximumlength measurementtechniques employed were to detectthe change in thesepa- sequencegenerated by the noisereduction system itself. Performance re- rationdistance acoustically of twoclosely spaced microphones placed in sultsare given for this system in a testsetup. Some issues regarding practi- theimmediate vicinity of thevelocity measuring probe and to determine cal implementationof thisalgorithm in fixedpoint arithmetic will alsobe theaccuracy of soundintensity measurements bycomparing with sound- discussed. pressurelevel measurements made in the farfieldof an acousticsource inside an anechoic room. The results show that there are noticeable changesin theacoustic field in thefrequency range 3000--6000 Hz dueto the insertionof the prnbe,and the scattering/diffractioneffects tend to 2:20 increasewith increase in theangle of incidenceof thesound waves, but do notaffect in anysignificant way the accuracy of soundintensity measure- QQ4. Accurate estimation of amplitude, frequency, and phase of ments.Measurement accuracies of theprobe are within 1 dB up to 5000 harmonic signal componentsusing an FFr. John C. Burgess Hz, whichincludes other sources of errorsassociated with signal process- (Department of Mechanical Engineering,University of Hawaii, ingsuch as errors inherent with FFT techniques. Honolulu, HI 98622) The signalanalyzed using an FFT is seldomidentical to thetruc sig- nal.This can result in inaccurateestimates of amplitude,frequency, and phaseof harmonicsignals. Using an extensionof a methoddescribed 1:50 earlier[J. C. Burgess,"On digitalspectral analysis of periodic signals," J. Acoust.Soc. Am. $8, 556-567 (1975) ], it is possibleto estimatethe true QQ2. Useof soundintensity for determinationof air-movingdevice nolse amplitude,frequency, and phase of a harmonicsignal with good accuracy. emission.A. C. Balant and George C. Maling, Jr. (Data Systems For example,acoustic intensity measurements may require determining Division,International Business Machines Corporation, Poughkeepsie, NY 12602) thephase difference between two harmonic signals of thesame amplitude andfrequency. Using the methoddescribed, this phase difference often AmericanNational Standardson soundpower determination via canbe estimatedwith an errorless than 0.01%. The methodrequires use soundintensity and on measurement of noise from air-moving devices are of an optimumdata windowthat canbc expressedas a Fourierseries at an advancedstage of development.The latter proposedstandard re- havingonly a few nonzerocoefficients. quiresa specialtest facility such as a reverberationroom or a hemi-ane- choicroom, and does not permitthe useof soundintensity for determina- tion of noiseemission. In this paper, the resultsof soundpower determinationson air-movingdevices in a reverberationroom, a hemi- 2:35 anechoicroom, and in anordinary laboratory environment are presented. QQS. The measurementof soundby the simultaneoususe of severaltime The datain theordinary room were obtained using sound intensity meth- constants.Svein Arne Nordby,Steiner Bohn (Norwegian Electronics, ods.The accuracy of thedata in all threeenvironments isdiscussed. Many P.O. Box 24, N-3408, Tranby, Norway), and Richard J. Peppin of the "fieldindicators" which are recommended in the proposedsound (Scantek,Inc., 51 Monroe Street,Suite 1606,Rockville, MD 20850) intensitystandard have been calculated in orderto determineacceptable rangesfor theseindicators. The resultsindicate that determination of air- Nonstationary or transient soundsmeasured by a soundlevel meter movingdevice sound power in ordinaryrooms is adequate for manyengi- will producedifferent values of soundlevel, depending on the time con- neeringpurposes. Use of soundintensity for determinationof air-moving stant,or lackthereof, used. The techniqueof determiningexposure, the devicesound power should increase the amount of dataavailable on these so-called"doubling rate" will alsoresult in significantdifferences in read- sources. ings.Most investigationsof the effectsof the differenttime constantsand averagingrates on nonstationarysignals required several sound level me- tersor a predefinedsignal that wasrepeated for eachdifferent measure- ment methodso that the differencescould be compared.This paperpre- 2:05 senCsthe results of testsusing a soundlevel meter that can simultaneously measurewith varioustime constants and with several"averaging" tech- QQ3. Investigationinto the use of thefiltered-x LMS algorithmfor noise niquesallowing the operatorto measureLma x andLmi n using"fast," reduction in a duct. Peter L. Schuck (Physics Division, National "slow,""impulse" time constants and "peak" detection using a doubling ResearchCouncil, Montreal Road, Ottawa, Ontario KIA 0R6, Canada) rateof 3, 4, 5, 6, andimpulse, all on the samearbitrary source.
S92 d. Acoust.Sec. Am. Suppl. 1, VoL83, Spring1988 115thMeeting: Acoustical Society of America S92 2:50 Excess attenuation of sound over a 7500-ft distance was measured on 6 days to determineeffects of temperatureinversions and wind direction. QQ6. Computationalstudies of the diffractionintegral occurring in the The testswere conducted using one-third octaves of pink noiseat center MAE theory of sound propagationover hills and valleys. Sames frequenciesof 250, 500, and 1000 Hz. During temperatureinversions, A. Kearns, Ji-xun Zhou, Yves H. Berthclot, and Allan D. Pierce excessattenuation downwind was found to be significantlyreduced and (Schoolof MechanicalEngineering, Georgia Institute of Technology, negative.Results of thetests were used to predictsound levels that would Atlanta, GA 30332) be receivedat islandresidences and a bird sanctuaryunder a proposed An importantclass of problemspertaining to outdoorsound propaga- industrial use of the mainland shoreline. tionis thatof thediffraction that occurs when the ground is neitherper- fectlyfiat norperfectly rigid. Such problems are encountered in thestudy of long-rangepropagation of soundover hills and valleys.It hasbeen shownpreviously [ J. Acoust. Soc. Am. Suppl.1 79, S30-S31 (1986) ] that the theoryof matchedasymptotic expansions allows one to expressthe 3:20 diffractedfield in termsof a complexintegral involving Airy functionsof QQ$. Model scale testing of outdoor noise propagationfrom a power complexargument. In somelimiting cases,the diffractionintegral re- plant. Robert A. Putnam (Gilbert Services,Inc., P.O. Box 1498, ducesto somecomputationally very efficientforms: an equationbased on Reading,PA 19603) geometricalacoustics in the illuminatedregion, a creepingwave series in the shadowzone, and a knife-edgeFresnel diffraction integral far behind A proposedcogenerating power plant of 70-MW capacityis to be the ridge.In the presentpaper, the transitionbetween these different re- locatedin anurban setting and will besubject to stringentenvironmental gimesis investigated numerically by computingthe general integral, and noiseemission criteria. A four-cellforced draft coolingtower is included particularattention is given to the matchingwith the creepingwaves in the plantdesign. Computer modeling of coolingtower noise propaga- seriessolution in the penumbraregion. Computational results are com- tion to affectedneighboring residences involved significant uncertainties paredwith data obtainedin laboratoryscaled experiments. [Work sup- dueto thecomplex shapes of powerplant building elements and existing portedby NASA LangleyResearch Center. ] surroundingbuildings. Without a finishedfacility in whichto perform full-scalefield tests,the next bestthing is a scalemodel test. This paper will discussthe specifies of the 1/20thscale test setup, materials, selection, instrumentation,test execution, some problems and some lessons learned. 3:05 The resultsof the testwill be givenin termsof the differencesin octave bandsound-pressure levels between an unobstructedfree-field value and QQ7. Sound attenuation in air over water on Puget Sound during the modeledin situvalue, at severalreceptor positions. The testenabled temperature inversion and noninversionconditions. Jan H. Hauge thespecification of more precise sound attenuating elements for thecool- (Towne, Richards & Chaudiere,Inc., 105 N. E. 56th Street,Seattle, WA ing towerfans and inlets in orderto complywith the governingregula- 98105) tions.
THURSDAY AFTERNOON, 19 MAY 1988 EAST BALLROOM B, 1:30 TO 5:00 P.M.
SessionRR. PhysicalAcoustics VII: Scattering
Kevin L. Williams, Chairman Code4120, Naval CoastalSystems Center, Panama City, Florida 34207-5000
ContributedPapers
1:30 1:45
RRI. A variationalprinciple for the scatteredwave. D.E. Freundand R. RR2. An analysis of partial wave scattering coefficientsand the, A. Farrell (The JohnsHopkins University Applied Physics Laboratory, interpretationof resonancesand backgroundin scatteringfrom elastic JohnsHopkins Road, Laurel, MD 20707) targets.M. F. Werby (NORDA NumericalModeling Division, NSTL, MS39529)andH. Oberall(PhysiesDepartment, Catholic University ot' Schwinger-typevariational principles are presentedfor the scattered America, Washington,DC :20064) wavein the caseof scalarwave scattering for an arbitraryincident field from an objectof arbitraryshape with eitherhomogeneous Dirichlet or Bistaticangular distributions and backscattered form functions can be, homogeneousNeumann boundary conditions. Designating the distance representedasa partialwave series where the angular contribution can be. from the scattererto the observerby r, then the resultsare variationally writtenin termsof sphericalharmonics or someother suitable angular invariantfor all valuesof r rangingfrom the surface of thescatterer to the function.It ispossible to represent elastic, rigid, and solid targets in sucha farfield.Explicit resultsare presentedfor the casewhen the scattereris a representation.The appropriatebackground (in partialwave space ) can sphereobeying homogeneous Dirichlet boundary conditions. Special at- be subtractedfrom someelastic target and a resonancecan be isolated. tentionis givento the selectionof the trial fieldsthat produceaccurate This is usuallydone by assuminga rigidor a softbackground for elastic: resultsover a broad frequencyrange. [Work supportedby the U.S. solidsor thin shellsand leads to a residualresponse that, underappropri- Navy. ] ate conditions,can manifesta resonance.Often, however,neither a rigid
S93 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring 1988 115thMeeting: Acoustical Society of America S93 nora softbackground is suitable. This can be established by examining the response(the form function) in an insonifiedacoustic field. The process- residualpartial wave coefficients, which, in theevent of a poorchoice of ing schemeuses the informationextracted from the characteristicsof the background,will leadto a mix of severalpartial waves at resonance.On form functionwithin a finite frequencyrange. In particular,it is found theother hand, for a goodbackground choice, a predictablemix of partial that the rateof phasechange with respectto frequencyin the form func- wavecoefficients will be observed.When a suitablebackground has been tionreveals three distinct types of reradiatedmodes, which are associated chosen,then the partial wave analysis allows one to characterizethe reso- with their underlyingphysical causes. Their separatedphysical param- nance. eterscan thus be determinedaccordingly. Examples with solidspheres constructedfrom differentmaterials are usedto illustratethe practical utilization of this study.
2:00
RR3. An analysisof the effectsof fluid loadingon the vibrationalmodes 2:45 of a submergedspherical shell. Gary S. Sammelmannand Roger H. Hackman (Physical Acoustics Branch, Naval Coastal Systems RR6. Identification of whispering gallery type waves on a finite Center, Panama City, FL 32407-5000) cylindrical shell using MIIR. S. K. Numrich (Code 5133, U.S. Naval Research Laboratory, Washington,DC 20375-5000), G. Maze, 1. Previousanalyses of the acousticscattering from submerged,spheri- cal elasticshells have relied on the subtractionof a "suitablydefined" L. Izbicki, and J. Ripoche (Laboratoire d'Eleetronique et backgroundcontribution to isolatescattering resonances of the shell.Un- d'Automatique,Universite du Havre,Place Robert Schuman, 76 610 Le Havre, France) fortunately,there is no universalagreement as to whatconstitutes a suit- ablebackground. In the presentwork, the acoustic scattering amplitude is Experimentswere conducted at Laboratoired'E!ectronique et d'Au- analyticallycontinued into the complex-kplane and the polesof thisam- tomatique(ULTRASONS) to confirmthe presence of whisperinggallery plitudeare directly located using a formof Cauchy's theorem. This proce- typewaves on a finitecylindrical shell bounded by hemispherical endcaps. durealleviates any difficulties associated with a backgroundchoice. Dis- The methodof isolationand identificationof resonances(MIIR) devel- persioncurves are generated for shellswith thicknesses ranging from 2% opedby the researchteam at ULTRASONS wasused to obtainthe bista- to 50% of the outer radius of the shell. This work verifies the bifureation tic modepatterns that conclusively identified the modes observed monos- of the firstsymmetric mode of a sphericalshell reported earlier [ L. H. taticallyby researchersat the Naval Research Laboratory. The radiation Green, R. H. Hackman, D. H. Trivett, and L. Flax, J. Acoust. Soc. Am. patternsfor theseveral low-order whispering gallery modes were identifi- Suppl. I 76, S65 (1984)], althoughthe detailsof the bifurcationdiffer ablewhen the cylindricalshell was insonified normal to its majoraxis. considerably.Our resultscall into question the whole concept of thesepa- Unequivocalidentification of higher-ordermodes was inhibited by the rationof polesinto background and elastic categories. presenceof wavestraveling along other circumferential paths.
2:15 3:00
RR4. GTD for backscatteringfrom elasticobjects in water:Phase of the RR7. Broadsideand end-onresonance scattering from submergedhigh- couplingcoefficient and a simplifiedsynthesis of the form function. P. aspectratio targets for severalelastic materials. H. 0berall(Physics L. Marston (Department of Physics,Washington State University, Department,Catholic University of America,Washington, DC 20064) Pullman, WA 99164) and K. L. Williams (Naval CoastalSystems and M. F. Werby (NORDA NumericalModeling Division, NSTL, MS Center,Panama City, FL 32407) 39529)
The geometricaltheory of diffraction(GTD) wasrecently extended Broadsideand end-onscattering from solidelastic spheroids com- to describesurface elastic wave (SEW) contributionsto scatteringfrom posedof aluminum,steel, and WC is examined.The spheroidsrange in fluid-loadedspheres and cylindersat highfrequencies [P. L. Marston,J. aspectratio from 4 to 10with a kL/2 rangeout to 26. A comparisonof the Acoust.Soc. Am. 83, 25-37 (1988) ]. The couplingof an SEW with the form functions for the different materials shows that the ratios of the kL/2 scatteredfield wasdescribed by a complexcoefficient G. That analysis valuesat resonancesare equal to the ratioof theRayleigh phase velocities gavesimple physical approximations: IG I = 8•r/•c/cswfor spheresand of the respectivematerials and that the end-onand broadside resonance ]G]= 8•r/•/(•rka)•/2for cylinders. The SEW angular attenuation coeffi- locationsare in accordance with the phase matching method of 'l•berall et cient(due to radiationdamping) and phase velocity are •' and Csw,while al., which is basedon the leaky Rayleightype wave interpretationof c denotesthe sound speed in thefluid, k = •o/c,and a isthe object's radius. resonances.Further, it is shown that the resonancelocations are only The presentwork givesapproximations for the phaseof G derivedin the sensitiveto the shearspeed and that it is possibleto generateresonances limit•',• 1 by assumingthat SEW contributions match those of resonance that occurend-on at all anglesincluding broadside, which is only consis- scatteringtheory based on a rigid background.The resultingphase of G is tent with the Rayleighwave interpretation of resonances. • 0 for solidspheres and • •r/4 for solidcylinders. The resultingapproxi- mation for spheresG•8•r/•c/csw yieldsa synthesisof the exact form functionsimilar in quality to previousones [K. L. Williams and P. L. Marston, J. Acoust. Soc. Am. 79, 1702-1708 (1986)] based on direct numericalevaluations of G. The resultssuggest that, given ff and Csw> c, 3:15 the SEW contributionmay be accurately approximated when ka• 1. The newsynthesis used a simpleapproximation for thespecular contribution. RRS. Acousticscattering from large-aspectratio, axisymmetricshells: [P. L. Marston was supportedby ONR.] II. Roger H. Hackmanand DouglasG. Todoroff (PhysicalAcoustics Branch,Naval CoastalSystems Center, Panama City, FL 32407-5000) PreviouslyJR. H. Hackmanand D. G. Todoroff,J. Acoust.Soc. Am. Suppl.I 82,S49 (1987) ], a comparisonwas presented of thepredictions 2:30 of the spheroidal-coordinatebased T matrixwith preliminarymeasure- RRS. Utilizing the form functions for determiningthe physical mentsof theacoustic scattering from a copper,prolate spheroidal shell at parametersof elasticsolid spheres.N. Yen and Louis R. Dragoncite axialincidence. This targethad an aspectratio ofL/D = 7 anda constant (PhysicalAcoustics Branch, Naval ResearchLaboratory, Washington, thicknessof 5% of the semiminoraxis. Here, more detailedmeasurements DC 20375-5000) arepresented and both the calculations and the measurementsto general incidenceare extended. The emphasisof thiswork is the physical analysis An approachis described to determinethe density and the longitudi- of theelastic excitations of thescatterer. The predictionsof the spheroi- nal and the shearvelocities of elasticsolid spheres based on their farfield dal-coordinatebased T matrixarc alsocompared with previouslypub-
$94 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America $94 lishedspherical-coordinate based T-matrix calculations forthin axisym- 4:15 metric shells. RR12. Boundedbeam transmissionacross water/sand interface. Part II--Experiment versustheory. K L. Williams and L. J. Satkowiak (NavalCoastal Systems Center, Code 4120, Panama City, FL 32407- 5O00) 3:30 Theexperiments described in the previous presentation allow the gen- RR9.Multiple scattering analysis for a targetin anoceanic waveguide. erationof contour plots of the pressure field within the sediment for differ- RogerH. Hackmanand Gary S. Sammelmann(Physical Acoustics enttypes of sendingtransducers at several grazing angles (defined here as Branch,Naval Coastal Systems Center, Panama City, FL 32407-5000) the anglebetween the interfaceand centralaxis of the sender).These contourplots are compared with theoretical predictions calculated using A multiplescattering analysis is presentedfor a targetin a range- the SAFARI model [H. Schmidtand F. B. Jensen,J. Acoust.Soc. Am. independentoceanic waveguide. The multiple scattering series isexplicit- 77,813-825 (1985) ]. Resultsindicate that the shear degree of freedom in ly summedand a solutionisobtained in closedform. The solution agrees withthat presented earlier [R. H. Hackmanand G. S.Sammelmann, J. the sedimentis unimportantunder the conditionsof the experiment. Acoust.Soc. Am. 80, 1447-1458(1986) ] for all casesthat werechecked. Agreementbetween SAFARI andexperimental pressure field contours is goodfor all sendersused in theexperiment. This suggests that nonlinear Themethod provides an attractive alternative to theprevious formalism effectsdue to parametric array truncation were not readily apparent in the in tworegards. First, this approach provides insight into the algebraic contourplots. Based on the experimentally determined in-sediment direc- structureof the solution. The individual terms in the final solution may be tionof propagation,the presence or absenceof beamdisplacement for simplyinterpreted in termsof singlescattering processes. And, second, eachtype of acousticsource will bediscussed. thereis less algebra involved when relatively simple waveguides are con- sidered.Numerical examples are given that illustrate the importance of contributionsinvolving rescattering among the target and the waveguide boundaries.
4:30
3:45 RR13. An iterative solution to acousticscattering by rigid bodies. Luise Schuetz, Joseph J. Shirron, and Ralph Kleinman (Naval RR10. Acousticresonance spectra of a finite cylindricalshell G, Maze, ResearchLaboratory, Washington, DC 20375) F, Lecroq,J.L. Izbicki,andJ. Ripochc(Laboratoircd'E]ectroniqucct d'Automatiquc,Univcrsit• Lc Havre, PlaceRobert Schuman,76610 Lc The scatteringproblem is formulatedas an integralequation over the Havre, France) scatteringsurface. A solutionto the equationis givenin a convergent Neumannseries where the convergencedepends on a parametera. The Previousstudies have explained acoustic scattering from an infinite boundaryelement method is used to obtainan approximation to thesolu- cylinderor a sphere.The scatteringdepends strongly on resonancesthat tion,and the resulting matrix equation is solved by an iterative technique,. are relatedto the propagationof circumferentialwaves when the acoustic Resultsare given for several different geometries and frequencies. A com- axisis perpendicularto the cylinderaxis or the propagationof guided parisonof solutiontime andaccuracy to directmethods such as Gaussian waveswhen the insonificationis oblique.The methodof isolationand eliminationis presented,showing the iterationto be moreeconomical in identificationofresonances (MIIR) developedin ourlaboratory is used mostcases of practicalinterest. Finally, thereis a shortdiscussion on to studyacoustic scattering from a finitecylindrical shell bounded by two determiningthe iteration parameter a andits effects on the rate of conver- planedisks. The acoustic beam diameter is larger than the target dimen- gence. sions.When the acoustic beam axis is perpendicular to the target axis, the resonancesobserved on the resonancespectrum are identicalto those observedwith an infinitecylindrical shell. When theacoustic beam axis is not perpendicularto the targetaxis, the resonancesare morenumerous and,for certainincidence angles, next resonances have the samemode. 4:45
RR14. Computationsof rigid body scatteringby long, slenderfinite cylindersusing the T-matrixmethod. Angie Sarkissian(Sachs/Freeman 4:00 Associates,1401 McCormick Drive, Landover, MD 20785)
RRIl. Bounded beam transmissionacross water/sand interface. Part l-- Waterman'sT-matrix algorithm requires the evaluationof matrixele- Experiment. L. J. Satkowiak and K. L. Williams (Naval Coastal mentsQ•k whose imaginary parts are difficult to evaluate for l < k forlong, SystemsCenter, Code 4120, Panama City, FL 34207-5000) slenderscatterers, at highfrequencies, because they involve surface inte- gralsof largeoscillatory functions. Machine precision required for nu- A seriesof measurementswas made to investigatethe effectof the mericalintegration of thesematrix elements increases as the frequencyis water/sedimentinterface on a propagatingacoustic signal. Under nearly increased.Therefore, the frequencyrange over which the methodcan be laboratoryconditions, data were collected mapping out thesound-pres- appliedis usuallylimited by theprecision of thecomputer used. Certain surefield in a homogeneoussand for a varietyof nonlinearand linear shapes,such as ellipsoids,have symmetric Q matrices;since numerical sources.The data were collected using the following sources: a truncated integrationof the difficultside of the matrix can be avoidedfor these parametricsource, a saturationlimited parametric source with effective shapes,the frequency range over which the method can be applied seems lengthless than the sender-interfacedistances, and a linear source,all to be limitedby machinememory alone. A differentapproach for the operatingat a frequencyof 20 kHz. Thedata were taken at grazingangles evaluationof the imaginarypart of Q will be shownfor finitecylinders thatranged from well-above to well-belowthe critical angle for thesedi- wherethe antisymmetric part of thematrix is written in a forminvolving ment.Based on thedata collected from the in-sedimentpressure field, surfaceintegrals that are more easily computed. By avoidingintegration contourswere obtained detailing the magnitude and direction of the pres- of largeoscillatory functions, this method overcomes the difficultiesasso- surewave in thesediment. These pressure field contours, as well as experi- ciatedwith longslender cylinders and enables high-frequency computa- mentalprocedures and analysis techniques, will be presented. A compari- tions.[Work supportedby Naval ResearchLaboratory, Washington, sonwith theoretical predictions will bemade in thefollowing paper. DC.]
S95 J.Acoust. Soc. Am, Suppl. 1, Vol, 83, Spring 1988 115thMeeting: Acoustical Society of America S95 THURSDAY AFTERNOON, 19 MAY 1988 GRAND BALLROOM B, 1:30 TO 4:45 P.M.
SessionSS. PhysiologicalAcoustics III: Structure and Function:The Ear Canal to the Auditory Nerve
Elizabeth C. Oesterle, Chairman Departmentof Otolaryngology,University of Washington,Seattle, Washington 98195
ContributedPapers
1:30 4.5 cm in standardbody length and 3 g in weight.They werefreely sus- pendedunderwater and anesthetizedduring the measurementprocess. SS1. Couplingbetween ear canaland eardrumin modelsystems. Michael Most of the responseswere recorded in vivo;however, some of the fishdid R. Stinson(Division of Physics,National ResearchCouncil, Ottawa notrecover from the anesthesia and may have died during the procedure. KIA 0R6, Canada) and Shyam M. Khanna (Department of The resultsare the first to verifythe existenceof a high-passfiltering Otolaryngology,Columbia University, New York, NY 10032) mechanismvia the Weberian ossicles, as previously postulated in theliter- This paperexamines a theoreticalmodel of soundpropagation in the ature. [Work supportedby ONR.] earcanal that accountsfor thedistributed acoustical load presented by the eardrum.This model is an extensionof our previousone-dimensional formulation IS. M. Khanna and M. R. Stinson,J. Acoust.Soc. Am. 77, 577-589( 1985) ], whichwas limited to rigid-walledcanals. The spatially varyingmotion of theeardrum enters as a drivingterm in theequation for the soundfield. In principle,this presentmodel can be combinedwith detailed,dynamical models of the tympanicmembrane. Experiments us- 2:15 ing modelcanals are underwayto testthe theory.In these,the soundfield is determinedwith the eardrumreplaced by eithera mechanicallydriven SS4. Model calculations of self- and external-tone interactions of pistonor a passive,locally reacting load. Preliminary results indicate that otoacousticemissions. Savithri Sivaramakrishnan,Glenis R. Long the theoretical formulation can describe the sound field in the cat ear (Departmentof A.udiologyand SpeechSciences, Purdue University, canalup to at least25 kHz andin the humanear canal up to at least15 WestLafayette, IN 47907), andArnold Tubis(Department of Physics, kHz. PurdueUniversity, West Lafayette,IN 47907)
Severaldifferent forms of nonlinear,active, cochlearmodels are used to simulatesome of the effectsof interactionsamong spontaneous oto- acousticemissions in the sameear [E. M. Burns, E. A. Strickland, A. 1:45 Tubis, and K. Jones,Hear. Res. 16, 271-278 (1984) ] and interactionsof emissionswith externaltones. The formerinclude mutual suppression, $S2. Determination of the reflection coefficient and the cross-sectional andintermodulation distortion products, while the latter include the sup- area function from high-frequencypressure measurementsin the ear pression,frequency shifting, and frequencylocking of spontaneousemis- canal. R. D. Rabbitt (Department of Mechanical Engineering, sions,and the lateholesof click and tone-burstevoked emissions. [Work WashingtonUniversity, St. Louis,MO 63130) supportedby NIH Grant No. NS22095.] At moderateto high frequencies,incoming and outgoingpressure wavescombine to formstanding waves in theear canal.The exactform of the pressuredistribution depends on th.egeometry of the ear canaland on the magnitude,and phase,of the reflectedwave relative to the incoming wave.An asymptotictheory describing this interaction is combinedwith pressuremeasurements in the ear canalin order to determinethe reflec- 2:30 tion coefficientat the eardrum, the cross-sectionalarea function of the ear canal,and the relative phase of thereflected wave. The theory is based on a SS5. Distribution of two major organ of Corti proteinsin different high-frequencymultiscale solution of the one-dimensionalhorn equation componentcell types.Peter Kraus, Jogy Varghese,Isolde Thaimann, and is shownto agreewell with experimentalmeasurements of standing RuedigerThaimann (Department of Otolaryngology,Washington wavepatterns and with thedistribution of phasein the earcanal. Since the University,St. Louis,MO 63110), and H.-P. Zenner(Department of absorptionof energyat theeardrum is determined, this method is particu- Otolaryngnlogy,University of Wiirzburg,D-8700 Wiirzburg,Federal larly well suitedto determinethe input to the ear at high frequencies. Republicof Germany) [Work supported,in part,by the WhitakerFoundation. ] Twohighly acidic, low molecular weight proteins, present in very high concentrationin theorgan ofCorti (OC) andtermed OCP-I andOCP-II [ I. Thaimannet aL, Arch. Otorhinolaryngol. 226, 123-128( 1980) ] have been previouslydescribed. The proteins do not contain significant amountsof carbohydrateand are solublein both high- and low-ionic 2:00 strengthbuffer. The distributionof theseunidentified proteins in different $S3. Measured frequencyresponses of the peripheralauditory or•an.• in celltypes of the OC hasnow been determined and has been compared with goldfish.Mardi C. Hastingsand PeterH. Rogers(George W. Woodruff the distribution of actin, tubulin, and calmodulin. Freeze-dried and fresh- Schoolof MechanicalEngineering, Georgia Institute of Technology, ly dissectedsubstructures of the OC of chinchillaand guineapig were Atlanta, GA. 30332-0405) used, respectively.Actin was found to be particularly abundantin the inner and outer pillar cells.In the Hensenand Deiterscells, OCP-I and The frequencyresponses of the primaryperipheral auditory organs in OCP-II predominateand calmodulin is present only in verysmall quanti- eightgoldfish (Carassius auratus) were measured using a recentlydevel- ties. In contrast,the outer hair ceilsexhibit a strongaccumulation of oped noninvasivevibration detection system basedon ultrasound and calmodulin,with relatively low levelsof OCP-I and OCP-II, actin and havinga spatialresolution of 0.28 mm[ M. Cox and P. H. Rogers,J. Vib. tubulin. In the neuroepitheliumof the sacculusOCP-I and OCP-II are Aeonst.Stress Reliability Design 109, 55-59 (1987) ]. The fishaveraged presentat about40 timeslower concentrations than in the OC.
ß96 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America $96 2:45 nataldevelopment in therat cochleahas now been determined. On the6th daypost parturn (pp), onlytrace amounts of thesubstances are present. SS6.Behavior of twomajor organ of Cortiproteins during development Theconcentration of OCP-I and OCP-II increasesmost rapidly between and maturationof the cochlea.Isolde Thalmann, Peter Kraus, days12 and 14 pp, and the proteins approach adult levels by day 18pp. $ogyVarghese, Thomas H. Comegys, RuedigerThaimann The sharpincrease of OCP-I andOCP-II occursat approximatelythe (Departmentof Otolaryngology,Washington University, St. Louis,MO sametime as the Sudden increase ofthe endolymphatic potential onday 13 63110),and H.-P. Zenner(Department of Otolaryngology,University pp [ S.K. Bosherand R. L. Warren,J. Physiol. 212, 739-761 ( 1971) ]. It is of Wfirzhurg,D-8700 Wiirzburg, Federal Republic of Germany) of interestthat a majorbasilar papilla-specific protein in thechick in- creasesin stepwith theonset of auditoryfunction during embryonic de- Twohighly acidic, low molecular weight proteins, present in very high velopment[J. C. Oberholtzeret al., Hear.Res. 23, 161-168(1986) ]. This concentrationin theorgan ofCorti (OC) andtermed OCP-I and OCP-II proteinis similar but not identical in its electrophoretic pattern to OCP-II. [ I. Thalmannet aL, Arch. Otorhinolaryngol. 226, 123-128 (1980) ] have Thebehavior of OCP-I andOCP-II will becompared with thepattern of beenpreviously described. The behavior of these proteins during the post- otherimportant proteins such as actin, tubulin, and calmodulin.
3:00-3:15 Break
3:15 3:45
S$7. Perturbationof cochlearmicrophonics by an intracellaiarelectrode SS9. Neurotransmitterand peptideneuromodulator candidates in hair- in the organ of Corti. J. J. Zwislocki and R. L. Smith (Institute for cell and nerve fractions of the saccular macula of the rainbow trout. SensoryResearch, Syracuse University, Syracuse,NY 13244-5290) Marian J. Drescher and Dennis G. Drescher (Laboratory of Bio- otology,Wayne State Medical School, 540 E. CanfieldAvenue, Detroit, Recordingthe receptorpotentials of cochlearinner and outer hair MI 48201 ) cellsin situhas become an essentialpart of cochlearresearch. To reachthe cells,the recordingelectrode has to traverseparts of the organof Corti. An epithelial sheetof 30 000-40 000 hair cells, detachedfrom the Suchintrusion may affectthe cochlearmicromechanics. This possibility basallamina of thetrout saccular macula, has been analyzed for primary- wastested by recordingscala-media CM in the secondcochlear turn in amineand N-acylated compounds by cation-exchangeHPLC. This hair- Mongoliangerbils. A smallhole was drilled in thecochlear bony capsule cell fraction (HC), greatlyreduced in supportingcells and neuralele- abovethe basilarmembrane, and an intracellularelectrode was made to ments,was simplerin primary-aminecomposition than the saccular traverseHensen's cells all the way to the OHCs. A secondelectrode nerve fraction (SN), which included afferent and efferent neurons and placedin theendolymph recorded CM amplitudeand phase as a function Schwanncells, The HC containedmore phosphoserineand phosphoeth- of soundfrequency. Our resultsindicate that an intracellularelectrode anolamine,or co-elutingcomponents, per microgramprotein than SN lodgedin theorgan of Corti canproduce substantial changes in CM tun- [cf. M. J. Drescherand D. G. Drescher,Comp. Blochem. Physiol. 86A, ing.This means that thecochlear micromechanics was perturbed. [Work 553-558 (1987)], and ten times more beta-alanine.Of the excitatory supportedby JayitsGrant NS 03950.] amino acidsin HC, glutamatewas present in highestconcentration, and wassignificantly elevated (p < 0.02), 2.6-fold,above levels determined for SN, consistentwith publishedefflux and releasestudies [M. J. Drescher, D. G. Drescher, and J. S. Hatfield, Brain Res. 417, 39-50 (1987) ]. Histi- dine-containingpeptides in millimolarconcentrations appeared uniquely partitioned:Camosine and homocarnosine, in a ratioof 10:1,respectively, 3:30 wereassociated with SN, while a third histidine-containingcomponent, SS8. Indirect evidencefor the absenceof electrotoniccoupling between tentativelyidentified as N-acetyl histidine,was associated primarily with hair andsupport cells in the mammaliancochlea. Elizabeth Oesterleand HC. Possibleinvolvement of histidine-containingcompounds as neuro- Peter Dallos (Departmentof Otolaryngology,RL-30, Universityof modulatorswill be discussed.[Work supportedby NIH Grant NS Washington,Seattle, WA 98195 and Auditory PhysiologyLaboratory 16166.] and Department of Neurobiologyand Physiology,Northwestern University,Evanston, IL 60201)
(3ap-junctionalcomplexes link goldfishand lizard hair cellswith adja- centsupporting cells [Hams, J. Neurocytol.9, 845-860 ( 1980); Nadol, 4:00 Mulroy, Goodenough,and Weiss,Am. J. Anat. 147, 281-302 (1976)]. SSI0. Quantitative light microscopiccomparison of cochlear nerve Lizardhair cellsand support cells appear to beelectrically coupled [Ba- morpholog• in Sprague-Dawley and Brown Norway rats. den-Kristensenand Weiss,Hear. Res.8, 295-315 (1982) ]. In mammals, Virginia Hoeffdingand Martin L. Feldman(Department of Anatomy, anatomicalevidence suggests the absenceof gap-junctionalcomplexes BostonUniversity School of Medicine,80 E. ConcordStreet, Boston, MA betweenhair cellsand support-cell neighbors, but suggestivemembrane 02118) specializationshave been observed [Oulley and Keese,J. Neurocytol.S, 479-507 (1976); Nadol,Ann. Otol.87, 70-80 ( 1978) ]. Horseradishper- Cochlearnerve morphology was studied in youngadult albino (Spra- oxidasemarking and intracellularrecording techniques were utilized to gue-Dawley)and pigmented (Brown Norway) rats.Following perfusion determinethe origin (s) of sound-inducedpotential changes in supporting of theanimals with mixed aldehydes, blocks including the cochlear nerves cellsof the organof Corti. Recordingswere made in the low-frequency wereremoved, embedded in Araldite,and sectionedin a planetransverse regionof the adult guinea-pigcochlea, an in oivopreparation. Response to the longitudinalaxis of the nerve.Analysis of the materialincluded characteristicsof supportcells to tonebursts at varioussound levels and countsof normal and degeneratingfibers and of giial cell nuclei,and frequencieswere compared with hair-cellresponses ( IHC andOHC) and measurementsof vascularityand of the nerve cross-sectionalarea. The potentialsrecorded in organ fluid spaces.Comparisons of the harmonic mediannumber of normal fibersin the Sprague-Dawleyrats was 21 216; contentof the responses,magnitude, and phaseof the fundamentalcom- in theBrown Norway rats it was20 186.There were no statistically signif- ponent,and the magnitudeof de-responsecomponents suggest a lack of icant differencesbetween the two strains in numbers of normal fibers, electrotoniccoupling between mammalian hair cellsand supportcells. degeneratingmyelin sheaths, or glial cell nuclei,or in the cross-sectional [Work supportedby NIH Grant NS-08635.] areasof thenerves. Vascular area density ( g'^D) wassignificantly higher
897 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115thMeeting: Acoustical Society of America S97 in theSprague-Dawley rats. The median ( V^D) in thatstrain was 0.0149, of a singleneuron. [ Work supportedby theDeafness Research Founda- while in the Brown Norway rats the median( V^i• ) was0.0099. [Work tion and NIH. ] supportedby NIH GrantsAG 00001and T32 NS07152.]
4:15 4:30
SSI1. Forwardmasking of the compoundaction potential: Thresholds for $$12. Auditory nervefiber responsespredict the discriminabilityof VOT the detectionof the NI peak. Evan M. Rclkin and Robert L. Smith syllables. Donal G. Sinex and Lynn P. McDonald (Los Alamos (Institute for SensoryResearch and Department of Bioengineering, National Laboratory, LS-1 M882, Los Alamos, NM 87545) SyracuseUniversity, Syracuse, NY 13244-5290) Responsesof chinchillaauditory nerve fibers to synthesizedsyllables A computerizedtwo-interval forced-choice method was used to mea- differing in VOT were obtained.A continuumheard as/da/-/ta/was surethe thresholdfor detectionof the N 1 peakof the compoundaction used.Many neuronsresponded with an increasein averagedischarge rate potential (CAP) recordedat the round window of the chinchillain re- or with a changein synchronizedresponse whose latency matched the sponseto a 23-msprobe tone. A 100-ms,pure-tone masker of equalfre- nominalstimulus VOT. Responselatency was treated as a randomvari- quencypreceded the probetone. Threshold shifts produced by themasker able.For eachVOT pair, a statisticanalogous to d' wascalculated from were measuredas the intensityof the maskerand the time delay between the distribution of the difference between two random variables. Thresh- the maskerand the probewere varied. The techniquesdiffer from prior oldsfor AVOT wereinterpolated from psychometricfunctions based on studiesin that forwardmasking of the compoundaction potential has this statistic.Thresholds were lowest for standardVOTs of 30-40 ms,but beenquantified as shifts in detectionthresholds rather than decrements in were highlydependent on the directionof changefrom the standard. the magnitudeof the N 1 peak.Results were compared to analogousmea- Thesecharacteristics were also typical of the chinchilla'spsychophysical suresfor singleneurons in the auditorynerve [E. M. Relkin and D. G. thresholdsfor/xVOT, reportedby Kuhl [ J. Acoust.Soc. Am. 70, 340-349 Peili,J. Acoust.Soc. Am. 82, 1679-1691(1987) ] andanalogous psycho- ( 1981) ]; in addition,the quantitativeagreement between neural and be- physicalmeasures of forwardmasking. Preliminary findings indicate that havioralthresholds was usuallyvery good.For this VOT continuum, forwardmasking of the CAP moreclosely corresponds to that observed discriminabilityreflects the precision of stimuluscoding at thelevel of the psychophysicallythandoes the forward masking observed inthe response auditorynerve. [Work supportedby NINCDS.]
THURSDAY AFTERNOON, 19 MAY 1988 418/420 ROOM, 1:30 P.M.
Meeting of Accredited Standards Committee S3 on Bioacoustics
L. A. Wilber, Chairman S3 422 Skokie Boulevard, Wilrnette, Illinois 60091 StandardsCommittee S3 on Bioacoustics.The currentstatus of standardsunder preparation will bediscussed. In additionto thosetopics of interestincluding hearing conservation, noise, dosimeters, hearing aids, etc., considerationwill begiven to newstandards that might beneeded over the nextfew years.Open discussion of Committeereports is encouraged. The international activities in ISO/TC 43 Acoustics, IEC/TC 29 Electroacoustics,IEC/TC 87 Ultrasonics, ISO/TC 108/SC4 and ISO/TC 94/SC12 will be discussed.The Chairs of the TAGs for ISO/TC 43 (H. E. yon Gierke),IEC/TC 29 (V. Nedzelnitsky),IEC/TC 87 (P. D. Edmonds),ISO/TC 108/SC4 and ISO/TC 94/ SC12 (H. E. yon Gierke) will report on currentactivities of theseTechnical Committees or Subcommittees.
S98 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spdng1988 115th Meeting:Acoustical Society of America $98 THURSDAY AFTERNOON, 19 MAY 1988 EAST BALLROOM A, 2:00 TO 4:45 P.M.
Session TT. Structural Acoustics and Vibration V: Characterization of Source Structure Interaction and Radiation Fields
Sabih I. Hayek, Chairman Departmentof EngineeringScience and Mechanics,Applied Research Laboratory, The PennsylvaniaState University, UniversityPark, Pennsylvania16802
ContributedPapers
2:00 2:30 l'fl. The Poyntingvector and source identification for thin platesand TT3. Responseof a vander Pol oscillatorto modulatedsinusoidal input. shells. Earl G. Williams, Phillip B. Abraham (Naval Research Huw G. Davies and Dennis Nandlall (Departmentof Mechanical Laboratory,Code 5137, Washington,DC 20375-5000), and Anthony Engineering,University of NewBrunswick, Fredericton, New Brunswick J. Romano (Sachs/Freeman Associates,Inc., 1401 McCormick Drive, E3B 5A3, Canada) Landover, MD 20784) The responseof a van der Pol oscillatorwith nonlineardamping and ThePoynting vector I, wellknown in thethree-dimensional theory of an externalperiodically modulated sinusoidal input is analyzedby using elasticity,represents the three-dimensionalpower flow in a solid.It is the methodof multipletime scales.The modulationappears to provide shownhow I canbe usedto derivethe powerflow in a thin flat plate,and a extendedstability. At resonance,the responseenvelope is highlynonlin- thincylindrical shell. The divergenceof I providesthe rateof changeof ear. Away from resonancethe modulationcauses a periodicswitching on totalenergy density (the HamiltonJandensity, •H/Ot) in the shell.Its and off of a naturalfrequency limit cyclesuperimposed on the excitation time averagemust vanish for steady-stateconditions. This leadsto the frequencyresponse. The naturalfrequency component arises through a conclusionthat power flow in and out of the structureis given by parametricexcitation. Extensive numerical simulations confirm the ana- H = • (I•D)•' h dl,a lineintegral (over the shell surface) of thein-plane, lytic results.Stability studies and the application to randommodulations averageintensity in theshell. If theshell is point driven, then H islarge at arealso discussed: [Work supportedby NSERC Canada.] that point,and can be usedto locatethe positionand strengthof the mechanicalsource. This provides an excellenttool for mechanicalsource identification.These concepts will be demonstratedand verifiedusing syntheticand real data on point driven cylindrical shells in water.In the syntheticdata, the full displacementvector (u,0,tv) is specified through- 2:45 out the shell.In the real data, measurementsof the radial component(tv) TY4. Effects of intensity modulationson the powerspectra of random of shellmotion are used.In the latter case(only tv known), it will be processes.William D. Mark (Physical Sciences Division, BBN shownhow it ispossible, however, to geta goodestimate of I1, andthus LaboratoriesInc., 10 Moulton Street,Cambridge, MA 02238} locate the mechanical source. Anintensity-modulated random process {x (t} } isdefined as the prod- uct of a deterministicmodulating function a(t) or modulatingprocess {a(t) ) anda stationary-modulatedprocess {z(t)) thatis statistically in- dependentof {a(t)}. Generalexpressions for theinstantaneous power spectraof intensity-modulatedprocesses are presentedfor variousclasses of modulatingfunctions and processes. A seriesexpansion of the instanta- neous power spectrumof intensity-modulatedprocesses is presented 2:15 which hasfor its first term a well-knownlocally stationaryspectrum ap- TT2. Uncertainty bounds for energy of imperfectly characterized proximation.This expansionis especiallyuseful when the fluctuation vibrationfields. Jean-Louis Guyader '• (Lab. Vibrations-Acoustique scalesTo of the modulatingfunctions are large in comparisonwith the InstitutNational des Sciences Appliqu6es de Lyon, 69621Villeurbanne, fluctuationscales T: of the modulatedprocesses. The expansioncan be France) interpretedas an asymptoticseries in the parameterTo/T z. For given scalesTO and Tz, it isshown that modulating processes {a(t) ) possessing Uncertainty boundsof vibration energy for viscoelasticcontinuous no firstderivative have a substantiallylarger effect on the powerspectrum media are given.They are related to a residualfunctional measuringthe of modulatedprocesses {z(t)) than modulatingprocesses possessing a error of characterizationassociated with an approximatesolution and a first derivative.Examples illustrating various aspects of the theory are factorthat can amplify or attenuatethe uncertainty bounds, depending on provided. the drivingfrequency. (That factorconsiderably amplifies the uncertain- ty whenthe structureis excitednear resonances. ) Assuming a statistical distribution of the error of characterization of the vibration field, over the modesof the structure,allows a considerablereduction of the energy uncertaintybounds. The caseof Gaussiandistribution is presentedand 3:00 analyzed for variousmean valuesand standard deviations.The bounds •fS. The design,construction, and testing of a miniatureshaker and force canbe usedas a criterionof convergenceof modalexpansion. The caseof gauge.H. Davis, B. Watters,and T. Graham (BBN LaboratoriesInc., theacoustical response of a roomis presented, and the influence of thesize Union Station, New London, CT 06320) and type, of the setsof modestaken into accountin calculations,is ana- lyzed.a• Actuallyon sabbaticalleave at Grouped'Acoustique de l'Uni- BBN has designed,fabricated, and tested a miniature combined versit6de Sherbrooke,Sherbrooke, Qu6bec J 1K 2R 1, Canada. shakerand forcegauge for mechanicalimpedance testing of light struc-
S99 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S99 ture(metal plating of theorder of 0.030in., for example).The design is pedances.The influenceof fluid loadingof the shellson the resonant basicallya Wileoxonminiature accelerometer to be drivenelectrically scatteringfrom the multiple-shell structure is addressed. Transition from throughthe signalcables, modified with a heavierreaction mass and a pressureboundary conditions to velocityboundary conditions is dis- substitutionofPZT 4 for theoriginal PZT 5A. The shakeris attached to a cussedand demonstratedwith numericalresults. Finally, effectsof com- stingerwire that supports the gravity load of the shaker while isolating the pliantcoatings on the shellsare examined. [Work supported,in part, by testobject from the rotary impedance of theshaker itself. The stinger wire the National ScienceFoundation. ] is attachedto a speciallyfabricated force gauge that is gluedto the test objectto minimizethe massbelow the forcegauge. The forcegauge is composedof two PZT 5A crystalsmounted back to back.The shaker designwill bedescribed and results presented showing that the shaker can 4:00 applyup to 150mN forcewith a sensitivityof 1 mN/V. Theforce gauge sensitivityis 350mV/N andthe frequency range is 2.6-6.6 kHz. By start- Tr9, Applicationof diffraction tomographyto acousticintensity ingout with an accelerometer design, BBN was able to producea working measurement.Woon S. Gan (Acoustical ServicesPte Ltd., 29 Telok miniature shaker in a minimum amount of time. Ayer Street,Singapore 0104, Republic of Singapore) Diffractiontomography is appliedto measureacoustic intensity in three-dimensionalspace. With Rytov'sapproximation, the homogeneous Helmholtzequation is obtained.The backpropagationalgorithm is used 3:15 to reconstructthe scatteredfield. Taking the specialease of incidentplane waves,the image kernel is expressed in convolutionform, and reconstruc- TT6. Randomresponse of a Duffing oscillator:The probability density tion becomesa deconvolutionprocess. An experimentalsetup is pro- function.Qiang Liu and Huw G. Davies (Department of Mechanical posed.The objectis a vibratingplate driven at its center.The acoustic Engineering,University of New Brunswick,Fredericton, New Brunswick pressureis measured from various angles. The pressuremeasurement acts E3B 5A3, Canada) as a diffractiontomogram. The particlevelocity is reconstructed.The A Duffingoscillator excited by narrow-bandrandom noise can exhibit structuralintensity (SI) is then computedusing the finite difference multivaluedbehavior, with randomjumps betweentwo pseudostable methodin termsof particlevelocity. A cross-sectionimage of the SI is states.In thispaper, an approximateform of the probabilitydensity func- obtained,which is an advantageover the nearfieldacoustic holography tion of the envelopeof the responseis obtained.State equations for the methodofL D. Maynard, E.G. Williams,and Y. Lec [J. Acoust.Soc. responseenvelope with the randomexcitation envelope as input are ob- Am. 78, 1395-1413 (1985) ], whichcan only givetbe acousticintensity as tainedby multipletime-scale analysis. The joint pdf of responseenvelope contoursand three-dimensional plots. From the cross-sectionimage, one and input envelopecan be foundapproximately for certainranges of the will beable to clarifythe outstanding tbeorctical problems of activeinten- systemparameters. The responseenvelope pdf showstwo maximain the sity and reactiveintensity. caseswhere the responseis multiple valued.Numerical simulationscon- firm the results.[Work supportedby NSERC Canada.]
4:15
3:30 Tr10. Diagnosticspectra ca•alogs.--A powerful means for predictionof faultsin machinery.M. Moustafa(Faculty of Engineering,Alexandria •r7. Recommended seismic zonation for Hawaii. W. M. Adams, A. University, Alexandria, Egypt), M. Shabara, A. El-Bahloul, and S. Furumoto, and E. Herreru-Bervera (HIG-410 A, University of M. Fanny (Faculty of Engineering, El-Mansoura University, EI- Hawaii, Honolulu, HI 9682:•) Mansoura,Egypt)
The stateof Hawaii hasboth volcanicand tectonicearthquakes: The A well-designedjoint will producelow levels of inherentvibration. As volcanicearthquakes are confinedto the Big Islandof Hawaii with the thejoint wears,however, clearance between mating components increases tectonicearthquakes occurring at all the majorislands, from Hawaii to andthis is reflectedin an increasein thevibration energy. As thisvibration Kauai. The historicalseismicity is relatively high, as might be expected energy is dissipated through the mechanism links, it excites fromthe presence of activevolcanoes and the transection of the Hawaiian resonancesand exerts extra dynamic loads on joints. It is of greatinterest archipelagoby the MolokaiFracture Zone. For the past120 years, the to the maintenanceengineers to predictthe occurrence of excessiveclear- extremeseismic intensity varies island to island:Kauai, IV; Oahu,VII; ancein joints,regarding its locationand magnitude. The factthat vibra- Molokai,VIII; Maui, VIII; Lanai,IX; and,for Hawaii, XII [H. D. Wood, tionsignals carry much information relating to tberunning conditions of J. Bull. Seismol.Soc. Am. 4, 169-203 ( 1914) ]. This informationsuggests jointscan be usedto solvethis problem. In thisstudy, a mathematical that an appropriateseismic zonation for buildingdesign is: zone1 for modelis formulatedto simulatethe effectof clearancein joints on the Kauai and Niihau; zone2 for Oahu;zone 3 for the islandsof Maul, Lanai, vibrationspectrum of machinesand mechanisms.The equationsof mo- and Molokai; and zone4 for Hawaii. By estimatingthe lossdistribution tion of the mechanism frame and links are formulated and solved simulta- suitablefor a particularlocation, one can derive risk functions useful for neouslyusing the Rungc-Kuttamethod. In doingso, and with the aid of settinginsurance levels [C. A. Williamsand R. M. Heins,Risk Manage- fastFourier transform (FFT), the frequencyspectra of the framein the ment and Insurance (McGraw-Hill, New York, 1985) ]. horizontal,vertical, and angular directions are obtained. Tbe spectraob- tainedfor the machinecorresponding to variousfaults with differentloca- tions and magnitudescan be usedto constructvibration charts which formthe "diagnosticspectra catalag." This DSC canbe used to determine the propertimes for maintenanceand the preparationof spareparts to 3:45 avoid failure of the machines. TrS. Scattering of sound from concentric cylindrical shells. Adnan Akay (Department of Mechanical Engineering,Wayne State University,Detroit, MI 48202)
An analyticalmodel of soundscattering from concentric,coaxial thin 4:30 cylindricalshells is developed.The specificmodel for scatteringfrom TTll. Effect of backlash on vibration spectrum in gear boxes. three concentric shells is examined in detail, and it is shown that the M. Moustafa(Faculty of Engineering,Alexandria University, Egypt), resultscan beeasily extended for any numberof shells.In order to seethe M. Shabara, Y. Abo-Mosallam, and F. Al-Saeidy (Faculty of effects of the shells and the annular fluid media between them more clear- Engineering,E1-Mansoura University, Egypt) ly, the caseof scatteringfrom shellsof infinitelength with a commonaxis parallelto the wavefront of the incidentplane wave is considered.The An idealgear box would produce no vibration at all becauseall energy problemis modeledin generalterms using radiation and mechanicalira- wouldbe channeled into thework to bedone; but a well-designedgear box
S100 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115thMeeting: Acoustical Society of America SllX) will producelow levelsof inherentvibration. As the gearteeth wear, boxwas used to demonstratethis model. In thissimulation, the transmis- however,backlash between meshing teeth increases and this is reflected in sionshafts were treated as lumped parameter systems. The elasticity c,f anincrease in the vibration energy. As the vibration energy isdissipated shafts,bearings, and gear teeth were also included. The grounded-chair throughthe gear box, it excitesresonances and exerts extra dynamic loads representationwas usedto obtainthe equationsof motionthat were ongear teeth. It isof great interest to the maintenance engineers topredict solvedusing the Runge-Kutta method. The vibration spectra of the gear the occurrenceof backlashbetween gear teeth. The factthat vibration box systemwere obtained using fast Fouriertransform (FFT). These signalscarry much information relating to runningconditions of gear spectracan be usedto constructa vibrationchart which, in turn,can b. • boxescan be usedto solvethis problem. In thisstudy, a mathematical usedto determinethe proper times for maintenance ofgear boxes to avoid modelwas formulated to simulatethe effect of backlashbetween gear failure of such units. teethon the vibration spectrum ofgear boxes. A singlestage helical gear
THURSDAY AFTERNOON, 19 MAY 1988 WEST BALLROOM B, 3:00 TO 5:05 P.M.
SessionUU. UnderwaterAcoustics VII: SignalProcessing Methods
David J. Thomson, Chairman DefenceResearch Establishment Pacific, FMO Victoria,British Columbia V8W 2Y2, Canada
Chairman's Introductionre3:00
ContributedPapers
3:05 that cannotbe predictedfrom conventional(narrow-band) ambiguity' functiontheory. As is well known,the theorycan be modifiedto include.. UUI. Underwater active array for echo cancellation.Application to the effectsof Dopplerdistortion on largeTW-product signals by correlat- reciprocal measurements.Gilles Chatel (Soci6t6d'Etudes et de Contr61e ing the receiw:dsignal against a referencethat is a time-compressedver- en Acoustiqueet Vibration, 6, Bd. de Brazza, 13008Marseille, France), sionof the transmittedsignal. In thispaper, the effectsof multipath(or' Thierry Rohan (Centred'Etude et de Recherchespour la Discr6tion targethighlight structure) and Doppleron the correlationprocess for' Acoustiquedes Nayires, Cerdan, Dcan Toulon, 83800 Toulon Naval, rectangularweighted large TW-product LFM waveformsare examined. France),Jean-Pierre Pasqualini (Consulting Engineer for Cerdan,Dcan Whenthe received echo contains no multipath, the correlator peak output Toulon,83800ToulonNaval, France),andAlain Route (Laboratoirede is a maximumfor the referencechannel that is closestin Dopplerto the: M6caniqueet d'Acoustique,CNRS, 13009MarseiHe, France) target.However, in a multipathenvironment, the correlatoroutput peak Reciprocalmeasurement of mechanical-acousticaltransfer functions doesnot generallyoccur at the correctDoppler referencechannel. The: is oftenused. If the radiatingstructure is immersedin shallowwater, the spreadingof the cross-ambiguityfunction as a resultof Dopplermismatch problemof parasiticreflections arises. A newmethod has therefore been andmultipath is calculated in termsof thedelay-Doppler coordinates and developed,based on theuse of a reciprocalarray of acousticsources. This magnitudeof the extrema.It is shownthat, undermultipath conditions, method allows real-time measurement of free field structure radiation on correctestimates of targetparameters can be recoveredthrough integra- a wide frequencyrange. Four acousticsources are monitoredso as to tion of the cotrelatoroutput. createwithin an observationvolume an acousticfield equivalent to the onegenerated by a monopolarsource placed in the freefield. The excita- tion signalsare calculatedby a stationaritymethod, taking into account 3:35 the intrinsiccharacteristics of the environment.First, testscarried out in shallowwater enabled us to checkthat the acousticpressure field mea- UU3. Effects of correlated noise on broadband conventional and modal suredwas the one required. Second, this method was applied to determine matched-field processing. G. B. Smith, D. R. De1 Balzo, and by reciprocitythe free field mechanical-acoustical transfer functions of an C. Feuillade( NavalOcean Research and Development Activity, NSTL, immersedcylinder. The resultswere compared to measurementsmade in MS 39529) the free field for the same structure. In a recentpaper [J. Acoust.Soc. Am. Suppl. 1 82, S73 (1987) ], the effectof realisticambient noise on narrow-bandmatched-field processing in both hydrophonespace and modespace was examined. A significant resultof that sr.udy was that correlatednoise degraded the performance of 3:20 the matched-fieldprocessor in hydrophonespace much more than it did UU2. Delay-Dopplerresolution performance of large time-bandwidth in modespace. In thispaper, the previouswork is extended to broadband productlinear FM signalsin a multipath ocean environment.Jean- in bothhydrophone space and modespace. Using computer simulations, Pierre Herinand (SACLANT ASW Research Centre, 1-19026 La the performanceof broadband conventionaland broadband modal Spezia, Italy) and William L Roderick (Naval Underwater Systems matched-fieldprocessing is comparedat variousbandwidths and various Center, New London, CT 06320) mixturesof correlatedand whitecomponents in the ambientnoise field. Sincethe averaging of correlated noise (which causes the modal processor Activesonar systems that transmit large time-bandwidth (TW) prod- to performbetter than the conventionalprocessor in realisticnoise fields) uct linear frequency-modulated(LFM) waveformsand receiveechoes dependson the numberof propagatingmodes, the effectof varyingthe from targetsof unknownspeed can sufferconsiderable correlation losses centerfrequency of theseprocessors is alsoexamined.
S101 J. Acoust,Sec. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S101 3:50 ofspatiallycorrelated versus spatially uncorrelated noise fields and choice ofarray configuration onsignal detection. Varying amounts of homogen- UU4. Matched-field processingand the effect of realistic noise eouswhite noise and spatially correlated noise as calculated by a normal correlation models. Cedric A. Zala (Barrodale Computing Services, mode noise model wcrc combined with the field due to a submerged Suite200, 1677Poplar Avenue, Victoria, British Columbia VSP 4K5, sourceto producesimulated phone space covariance matrices for a range Canada), John M. Ozard (Defence Researchœstablishment Pacific, of array configurationparameters. A phone-to-modespacing mapping FMO, Victoria, British Columbia V0S lB0, Canada), and was then usedto obtainthe correspondingcross-amplitude correlation Fansto Milinazzo (Royal Roads Military College, FMO, Victoria, matrices.The resultsof both conventionaland MLM processingwere British Columbia V0S lB0, Canada) analyzedand the trade-off between the number of phonesand percent of In simulatingacoustic source localization by matched-fieldprocess- the watercolumn spand was investigated. ing, it is desirableto usea realisticambient noise model. Two general classesof noisemodel were considered:a planar isotropicnoise source distributionand a line sourcedistribution. The formerwas implemented basedon the work of Kupermanand ]ngenito[J. Acoust.Soc. Am. 6?, 1988-1996(]980) ]. Their expressionsfor the correlationintegrals were adaptedto computethe correlation function for bothsimple and realistic propagationmodels, through the use of analytical and numerical Green's 4:35 functions,respectively. The line source model involved the assumption of UU7. A new method of ocean acoustic tomographym"Phsse uncorrelatednoise sources along an infiniteline and wasdesigned to mod- el a shearzone in a shallowice-covered ocean. Using the corresponding tomography."E. C. Shang (Wave PropagationLaboratory, NOAA, Green'sfunctions for a waveguide,the line modelcorrelation function Boulder, CO 80303) couldbe reducedto a summationover all modepairs, each term of which The ray traveltime has been used as the basic measurement for inver- involvedan integrationover wavenumber.In this paper,the line source sionof thesound speed structure in theconventional ocean acoustic tomo- correlationmodel is presentedand correlationfunctions given by both graphy.There are caseswhere the ray theoryis neitherapplicable nor noisemodels are illustrated. The effectof usingthese modcls on thelocal- convenient,especially for lower frequencyand axial propagationin the izationof a discretesource by a gridsearch-optimization procedurc is then oceansound channel. In thepresent paper, a newmethod of oceanacous- examined. tic tomographyhas been developed; it is a full-waveapproach based on the adiabaticnormal mode theory. Instead of the ray travel time, the modal phasedifference perturbations are usedfor inversion--wecall it "phase tomography."The methodis suitablefor lower frequency,and it is also 4.'O5 suitablefor axial propagation;moreover, the range-dependentsound- speedstructure can be obtainedfrom the verticalslice inversions. The UU5. The sensitivity of experimental, shallow-water, matched-field mathematicalderivation will bcpresented and some possible ways of im- localization estimates to variation of environmental parameters. plementationwill bc discussed. C. Feuillade, W. A. Kinney, and D. R. Del Balzo (Naval Ocean Researchand Development Activity, NSTL, MS 39529) In two recentpapers [ J. Acoust.Soc. Am. Suppl.1 80, SI 13 (1986) ], the results of simulation studies to determine the effects of environmental mismatchon the robustnessof matched-fieldprocessors in shallowwater werestudied. In this paper,the knowledgegained from thosestudies is 4:50 appliedto the interpretationof a verticalarray shallow-water,matched- fieldexperiment off PanamaCity, FL. How theability to localizein range UU8. Remote sensingof ocean flows with spatial filters. Gregory and depth is dependentupon the followingfactors is discussed:water B. Crawford (CIRES, University of Colorado/NOAA, Boulder, CO depth,sound-speed discontinuity at the water/sedimentinterface, and 80309), Richard J. Lataitis, Stevcn F. Clifford (Wave Propagation sedimentdensity and attenuation. Also discussed will bethe factthat the Laboratory,NOAA./ERLf0VPL, R/E/WP, 325 Broadway,Boulder, resultsindicate: (a) the limitationsof single-frequency,time*indepen- CO 80303), and David M. Farmer (Institute of OceanSciences, 9860 W. dent,matched-field processing and (b) the needfor broadbandand time- SaanichRoad, Sidney,British Columbia VSL 4B2, Canada) dependenttechniques. Thccomplicated scintillation pattern observed when an acoustic wave propagatesthrough refractive indcx irregularities is ducto the "projec- tion" of all the refractiveeddies at eachpath position onto the receiving plane.A currenttransverse to theacoustic path will adveetthe eddies and 4:20 producca drift in the scintillationpattern. A singletransmitter and two receiverscan be usedto monitor the drift and infer a path-averagedflow UU6. Noise and army configurationconsiderations for matched-field speed.Much morelocalized measurements of the transversecurrent can processingin phoneand mode space. G.M. FrichterandD. R. DelBalzo bemade with multiplesources and receivers that spatially filter the scintil- (Naval OceanResearch and DevelopmentActivity, NSTL, MS 39529) lationpattern. Such spatial filter systems can be used to profilethe trans- The conditionsunder which mode space matched-field processing of- verse current. Recent results obtained with small transmitter and receiver fersperformance gains relative to phonespace matched-field processing arraysmounted across a well-mixed,turbulent tidal channelare present- areinvestigated. Particular emphasis is placedon determiningthe effects ed.
$102 J. Acoust.Sec. Am. Suppl.1, Vol. 83, Spring1988 115th Meeting:Acoustical Society of America $102 THURSDAY AFTERNOON, 19MAY 1988 418/420ROOM, 3:00P.M.
Meeting of AccreditedStandards Committee S1 on Acoustics
D. Johnson,Chairman S1 Larson-DavisLaboratories, 280 South Main, PleasantGrove, Utah 84062
StandardsCommittee SI onAcoustics. Working group Chairs will report on their progress in thepreparation of standards,methods of measurementand testing, and terminology in physicalacoustics, electroacoustics, sonics,ultrasonics, and underwater sound. Work in progressincludes measurement of noisesources, noise dosimeters,integrating sound-level meters, and revision and extension of soundlevel meter specifications. Opendiscussion of committeereports is encouraged. The internationalactivities in ISO/TC 43 Acoustics,IEC/TC 29 Electroacoustics,and IEC/TC 87 Ultrason- icswill be discussed. The Chairs of therespective U.S. TechnicalAdvisory. Groups for ISO/TC 43 (H. E. yon Gierke),IEC/TC 29 (V. Nedzelnitsky), and IEC/TC 87 (P. D. Edmonds)will report on current activities of these Technical Committees.
S103 J. Acoust.Sec. Am. Suppl. 1, Vol.83, Spring 1988 115thMeeting: Acoustical Society of America S103 FRIDAY MORNING, 20 MAY 1988 EAST BALLROOM B, 8:00TO 11:00A.M.
SessionVV. EngineeringAcoustics V: GeneralEngineering Acoustics
StevenC. Thompson,Chairman Acousticsand Sensor Systems, Ocean Systems Division, GouM Inc., 18901Euclid Avenue, Cleveland, Ohio 44117
ContributedPapers
8:00 and elasticscreens in water.The casesof the infinitestrip (2-D space)and ofthe rectangular plane screen (3-D space) are considered. In both cases, VVl. The attentuationof dissipativeflow duct mufflerswith internal it is assumedthat the thicknessof the screenis smallcompared to the mean fluid flow in the absorbent.A. Cummings (Department of wavelength.The calculations are made up to ka = 10 (k, wavenumber;a, EngineeringDesign and Manufacture,University of Hull, Hull HU6 characteristicdimension). The resultsare comparedto experimentson 7RX, United Kingdom) softmaterial and steel screens, and to the infiniteplane model. It is shown In dissipativeflow duct mufflers incorporating layers of porousbulk thatin thisfrequency range, the scattering effects due to thefinite size of soundabsorbent, mean flow within the poresof the materialis inevitable thescreens are important, mainly for thesteel panel. becausethe porousmedium is exposedto pressuregradients in the duct flow. The acoustical effects of internal mean flow can be considerable, particularlyin the caseof internalcumbustion engine exhaust mufflers of the dissipativetype. Theoretical treatments are presentedhere, both to describethe effectof internalflow on the bulk acousticproperties of the porousmedium and to findthe effect of theabsorbent in situ,in theform 8:45 of the sound transmission loss of the silencer. The measured transmission lossof an experimentalsilencer is comparedto predicteddata, and good VV4. Ultrasonictomography in thefield of flow measurement. $•rg Wolf agreementbetween the two is obtained. The effects of meanfluid flow in (Institutf'tir Mess-und Regelungstechnik, Universitiit (TH) Karlsruhe, thecentral passage and internal fl0w in the absorbent areseparately dem- Postfach6980, D-7500 KarlsruheI, FederalRepublic of Germany) onstrated. Ultrasonictomography is knownas a nonintrusiveimaging method forstatic sound attenuating objects in fields such as medical diagnostics or nondestructivetesting. A newapplication of ultrasonictomography in chemicalengineering ispresented in thispaper, namely, the nonintrusive 8:15 imagingof thestatic fluctuating void fraction in a crosssection of a gas liquidflow. Here, the attenuation of parallelultrasonic rays is measured. VV2. Perturbations of the acoustic field inside a resonant cavity. The setof valuesobtained from this measurement result in a projectionof PhilippeHerzog and Michel Bruneau (Laboratoired'Acoustique, theflow. Now thespatial distribution of voidfraction can be reconstruct- Universit• du Maine, B.P. 535, 72017 Le Mans Cedex, France) edfrom a setof these projections, taken from various directions, bythe algorithmsof computerizedtomography [Proc. IEEE, 21(3) (1983)]. A theoreticaland experimentalstudy has been made of the acoustic Themethod was applied to a bubblecolumn as used in chemicalengineer- field insidea closedresonant cavity with two axesof resonanceunder the ing.To geta fastand nondisturbed measurement of theprojection data, a influenceof severalperturbations, like an imperfectgeometry, the pres- circulararray of 108piezoceramic transducers was directly integrated in enceof the transducers,and the effectsof the motionof the cavity (linear thepipe wall of thecolumn. The transducers are electronically switched andangular speeds and accelerations). The analyticalstudy is based on a andneed not be mechanically operated. The controlof thearray and the modal decompositionof the pressure,for Cartesianand cylindrical reconstructionalgorithms are implemented on a microcomputercoupled shapes.The influence of geometrical perturbations and the presence of the with an array processor. transducersare taken into account by a perturbationmethod applied to a matrixformulation of theHelmholtz-Huygens equation. The influenceof inertialterms is expressed in a similarform, but one which may be used for greatermode coupling. The experimental work has been done in theCar- tesiancase for a two-dimensionalcavity. The useof a swept-sinusoidal excitationwith synchronousdemodulation and a custom-madeacoustic antennapermitted us to obtainvery accurate measurements. Several ex- 9:00 amplesare shown, and the agreement with theory is quite good, leading to VVS. Scatteringfrom slots. Marian Smith (Lockheed Palo Alto discrepanciestypically less than 0.1 dB andthe detectionof modesin a ResearchLaboratory, 0/91-60, B/256, 3251Hanover Street, Palo Alto, dynamicrange of about50 dB. CA 94304)
The effect of slots in boundaries on scattered radiation is a classical problem,yet one which is still of considerableinterest. This paper presents an extensionof priorwork by the author[ J. SoundVib. (to bepublished 8:30 July 1988)], whichconsidered the effectof coveringonly the longwave- W3. Scatteringfrom thin elasticscreens. Christian Audoly (Groupe lengthapproximation, where tbe slotwidth wasmuch smaller than the d'Etudeset de Recherchede DetectionSous-Marine, Le Brusc,83140 Six- wavelengthof the incidentradiation. This paperpresents a moregeneral methodfor the calculationof scatteringfrom slots,which allows for solu- Fours, France) tionsfor largerslots, limited only by resonanceconsiderations and com- The characteristicsof acousticalmaterials are often determined by putersize. Examples of solutionsfor scatteringfrom slots in rigidwalled standardpanel tests. To improvelow-frequency measurements, the scat- plates,ducts, and cavities are presented. The methodreduces the scatter- teringeffects due to the finitesize of thepanels are investigated. A theo- ing problem,by meansof the Green'sfunction techniques, to an integral reticalmodel, based on the Helmholtz integralequation and its derivative or integrodifferentialequation. The resultingequation can be solvedby form,is usedto calculatethe scatteringof acousticwaves from rigid, soft, numerical methods.
S104 d. Acoust.Soc. Am. Suppl. 1, VoL83, Spring1988 115thMeeting: Acoustical Society of America S104 •15 by solvinga relativelysmall system of linear equations.Experimental measurementsof the networktransmission loss were madeusing a two- W6. Laser Doppleranemometry detection of hydroacousticparticle microphonetransfer function technique with random "white" noise and a velocity. P.S. Dubbleday and H. C. Schau (Underwater Sound dualchannel FFT analyzer.The resultantsound transmission loss was in ReferenceDetachment, Naval ReseaxchLaboratory, P.O. Box 568337, good agreementwith the predictedtransmission loss up to the cot-on Orlando, FL 32856-8337} frequencyof crossmodes within the straightpipe sections. The methodof LaserDoppler Anemometry (LDA) is appliedto the measurementof particle velocityin a hydroacousticfield. The field is createdin a USRD type GI9 or G40 calibrator.The LDA resultsare comparedwith the valuesfor the velocityobtained from the pressure gradientmeasured by a USRD standardhydrophone. The anemometeris operatedin forwardscattering mode, and alsoin backscatteringmode 9:45 throughthe useof a fiber-optictransducer. Seeding particles of varying materialand size are dispersed in theacoustic medium (water}. The influ- VVS. The use of numericaloptimization techniques in the designof enceof these parameters including the particle concentration on the signal acoustichorn loudspeakers.James McLean (TRW AntennaSystems qualityand detection threshold is discussed. [Work supportedby ONR.] Laborator},,Redondo Beach, CA 90277} and E. L. Hixson (The Universityof Texasat Austin,Austin, TX 78712) Theuse of numerical optimization techniques in thedesignof acoustic horn loudspeakersystems is considered.Conventional synthesis tech- 9:30 niques(i.e., stepped impcdance transformer ornonuniform line) are in- VV7. Transmissionloss of a piping network. D.C. Stredulinsky, adequatefor hornloudspeaker design due to the frequency-dependent natureof the radiativeload at themouth and the complex behavior of the A. Craggs, and M. G. Faulkner (Department of Mechanical Engineering,University of Alberta, Edmonton, Alberta T6G 2G8, driver.In thispaper, the hornis modeledas an n-sectioncommensurate Canada) steppedimpedance transformer terminated in a one-portnetwork that representsthe complex frequency-dependent radiation impedance. The Thispaper describes a numerical procedure developed to analyzepip- driveris modeled as a lineartwo-port transducer using measured data as ing andduct systems. The procedurecombines transfer matrix and finite describedin anearlier presentation [J. S. McLeanand E. L. Hixson,J. elementmethods and was used to predictthe transmissionloss of a piping Acoust.Soc. Am. Suppl.I 79, S90(1986) ]. The entirehorn/driver sys- network constructedof rigid-walled,square cross-section tubing. The temis optimized to exhibitequiripple frequency response using the char- networkconsisted of oneinput, one output, and several segments made up acteristicimpedances ofthe individual transformer sections asthe optimi- of straightpipes and 90 • elbows.These were interconnected atseveral "T" zationvariables. The system is optimizedas a singlyterminated filter in junctions.The junctionsand elbowswere treatedusing finite element orderthat it maybe driven by a Iow-impendancepower amplifier. The modelsthat tookinto account two-dimensional wave propagation. These optimizedhorn/driver system exhibits a bandpass response. This differs wereconnected to straightpipes in whichone-dimensional wave propaga- fromprevious horn designs which were high pass in nature.In theopti- tion was assumed. This led to a much reduced connective matrix for each mizeddesign, the bandwidth ofthe system may be increased byallowing a finite element model. The solution of the entire network was then obtained higherripple level and vice versa.
10.'00-10:15
Break
10:15 Radarwarning receivers (RWR) aresystems that alert pilots of mili- taryaircraft to imminentdanger of attackby missilesand gunfire. These •9. Influenceof groundreflection on measurementsinvolving bands of RWRs functionto: (a) interceptradar transmissions; (b) detectthe ra- noise.John M. Noble, Henry E. Bass,and Richard Raspet(National dar'spulsed modulations; (c) identifythe patterns of pulsesas emanating Centerfor PhysicalAcoustics, P.O. Box847, University, MS 38677) fromfriend or foe,and, finally, (d) presentinformation as to threattype It hasbeen shown [L. C. Sutherlandand H. E. Bass,J. Acoust.Soc. anddirection via alfa numerics and symbols on a videodisplay and, aural- Am. 66, 885-894 ( 1979}] that frequency-dependentatomospheric ab- ly, viaa heatset. The aural presentation termed "raw audio" is entirely an sorptioncan lead to a propagationloss for a bandof noisethat ismuch analogprocess. It isused to complement the video display as an aid in the differentfrom that for a puretone at bandcenter. In the presenceof a detectionand recognition of threats.In orderto decreaseclutter and dis- groundsurface, interference can also cause the sound amplitude to vary tractionto thepilot, modern RWRs incorporated computer analysis to rapidlywith frequency.When this occurs, the level measured for a pure assistin discerningthreats from the myriadof PRF patterus.With tonecan differ dramatically from that measuredfor a bandof noise.Accu- compterlogic, of course,the demodulated pulse data must first be con- ratetreatment of thisdifference requires integration over the bandpass of vertedto digitalform. This process, however, has eliminated the analog the fractionaloctave band filter used in the measurement.Example calcu- raw audioin modernRWRs in favorof"synthetic audio," reconstituted lationshave been performed for a typicalfilter. These examples form a fromdigital data. This paper discusses the differences between raw audio basisfor generalguidelines to beused when comparing theory to measure- and syntheticaudio and their advantagesand disadvantages.The paper ments. concludesthat computer-driven RWRs supplementedwith rawaudio ca- pabilityshow considerable promise for improveddetection and recogni- tion capabilityand should be expeditiouslyinvestigated.
10'.30 10:45 VVI0. Audio processingfor improveddetection and recognitionof threats in radar warning systems.Charles H. Wiseman (Lockheed ¾Vll. S•mand difference frequency generation in soundradiation from Missiles& SpaceCompany, Inc., AstronauticsDivision, Advanced a vibratingplanar boundary. Mosaad A. Foda (Faculty of Engineering, Marine Systems,P.O. Box 3504,Sunnyvale, CA 94088-3504) EI-MansouraUniversity, Egypt)
S105 J. Acoust.Sec. Am. Suppl. 1, Vol.83, Spdng1988 I 15thMeeting: Acoustical Society of America S105 An earlierstudy [J. H. Ginsberg,J. Acoust.Sec. Am. 69, 60-65 obtainedas a specialcase of thepresent analysis. Using the Fourier trans- ( 1981); 69, 929-936 ( 1981) ] of soundradiation from a vibratingplanar formin itsconventional complex function form to describethe linearized boundaryconsidered the caseof monochromaticexcitation. That work signalleads to fomulationof thesecond-order potential in termsof com- featuredperturbation analysis combining asymptotic integration and the plexfunctions. A coordinatestraining transformation describing the full renormalizationversion of the methodof strainedcoordinates. The pres- spectrum(propagating and evanescent) is deduced. Accordingly, uni- entanalysis initiates an extension of thosetechniques to thecase era two- formlyaccurate expressions for the acousticpressure and the velocity frequencyexcitation. These frequencies are considered arbitrary; hence, componentsare obtained. the resultsfor a parametricarray (closelyspaced frequencies) can be
FRIDAY MORNING, 20 MAY 1988 EAST BALLROOM A, 8:10 TO 11:30 A,M.
SessionWW. Educationin AcousticsIV and PhysicalAcoustics VIII: Computer-BasedTheoretical Studies in Acoustics
CharlesThompson, Chairman Departmentof ElectricalEngineering, University of Lowell,1 UniversityAvenue, Lowell, Massachusetts 01854
Chairman's Introduction•8:10
Invited Papers
8:15
WW1. MACSVMAautomated symbolic mathematics •oftw•re. Richard Petri (SymbolicsInc., Cambridge,MA 02139) Justas computers revolutionized numerical analysis, so symbolic mathematics software is revolutionizing symbolicmathematics computations, yielding major increases in speedaccuracy and modeling power in many applications.In addition to automating large and tedious computations, MACSYMA provides automated math- ematicalexpertise in manyareas, such as solution of simultaneous equations, integration, Laplace transforms, ordinarydifferential equations, and simplifications of expressions. The presentationswill focuson selected capabilitiesof MACSYMAthat are relevant for acoustics, especially: ( 1) symbolicvector and tensor calculus, (2) symbolicsolution of ordinarydifferential equations, (3) symbolicapproximation methods, (4) useof symbolicmath software to generatenumerical analysis code, (5) symbolicmatrix computations. A partial bibliographyof researchpapers in fluidmechanics reporting on work where MACSYMA was used will alsobe provided.Also discussed will be the role of symbolicmathematics software in technicaleducation.
8:45
WW2. SeratehpadII: A computeralgebra languageand system.Richard Jenks (T. J. Watson Research Center,Yorktown Heights,NY 10598) The ScratchpadII systemrepresents a new generation of systemsfor doingsymbolic mathematics, based on modernalgebra and abstractdata types.A largenumber of facilitiesare provided,for example:symbolic integration,"infinite" power series, differential operators, Cartesian tensors, and solutionof nonlinearsys- tems.Scratchpad II hasbeen designed from the outsetto be extendible.The systemintroduces a new data abstractionnotion, the "category," to expressintricate interrelationships between data types. The resultdesign permitsthe compilation of algorithms described in theirmost natural mathematical setting. The use of categor- iesguarantees user defined types and packages are compatible with eachother and with builtin facilities.This systemprovides a singlehigh-level language with an intepreterand compiler. The.language can be usedby the naive user for convenientinteractive mathematics calculations and by the advanceduser for the efficient implementationofalgorithms. Scratchpad II isbuilt on Lisp/VM andruns on IBM/370 classmainframes. An implementationof the systemon the RT/PC is expectedsoon.
S106 J. Acoust.Sec. Am. Suppl.1, Vol. 83, Spring1988 115th Meeting:Acoustical Society of America 8106 9:15
WW3. A surveyof FORTRANcode generation using MACSYMA. Stanly Steinberg(Department of Mathematics andStatistics, University of New Mexico,Albuquerque, NM 87106)and Patrick J. Roache(Eeodynamic• ResearchAssociates Inc., Albuquerque,NM 87106) This paperpresents the underlyingsymbolic manipulation techniques used by the authorsto produce FORTRANcode with theartificial intelligence code MACSYMA. The FORTRANcodes produced use finite differ- enccmethods to solvethe governing equations for numericalgrid generalion, including elliptic and variational methods,and to solvethe hosted equations, including fluid dynamics, heat transfer, and electro-statics. Consi- derationwill be givento writingthe hostedequations in strongconservation form in generalnonorthogonal coordinates;high order and conditional differencing schemes; colocated variables versus staggered grids; and codevalidation procedures.
9:45
WW4. Use of symbolicmanipulation in physicalacoustics. Martin ManIcy and Vinect Mehta (Electrical EngineeringDepartment, University of Lowell,I UniversityAvenue, Lowell, MA 01854) Thedevelopment of a mathematicaltheory in acoustics (or in anyother field } dependson the insight of the theoretician.However, the formulationof an insightinto an equationtierived from first principles involves muchtime spent in activitiesthat fall within the category of a symbolicmanipulation. To demonstratethe role symbolicmanipulation software can play in thesolution process, classical problems inacoustics will be select- ed.It isknown that the solution to theiincarized governing equations of an acoustic field can be considered as a superpositionofvorticity, thermal, and acoustic modes. However, these modes are coupled. The separation of thesemodes will bedemonstrated. Symbolic manipulation will be usedto explorehow the higher-orderterms inthe perturbation parameters, e,,and e•, affect the modal coupling. The issue of asymptotic degeneracy will alsobe addressed (the effects of varyingthe rates at whichthese parameters approach zero).
10:15
WWS. An evaluationof an intelligentcomputer service assistant. Paul W. Armstrongand Dan LaRue (Apollo Computer,Inc., Chelmsford,MA 01824) Thispresentation will discussthe development issues required for theimplementation of an intelligent assistantfor Apollo'scustomer service organization. An overviewof thedesign and requirement goals as well asthe integrationwith currentdata systems and input will be addressed.Maintenance and future enhance- meritswill be reviewedas well as a discussionof the userinterface and userprofiles.
ContributedPapers
10-A5 Two acousticscourses are offeredby the Naval AcademyPhysics Department.One is a physicsmajor elective course in physicalacoustics WW6. Applicationof symboliccomputation to singularperturbation with a weekly2-h lab;the other,a sonarcourse, is basicallya service problemsin physicalacoustics. Charles Thompson (Department of coursefor other departments. On accountof thelarge attendance for the ElectricalEngineering, University of Lowell, I UniversityAvenue, latter,and the limited space available, it hasnot been possible to makelab Lowell, MA 01854} worka routinepart of thecourse. Since it isbelieved that some lab experi- Problemsin fluid mechanics,and in turn in physicalacoustics, often enceis essential to good physics teaching, materials are being created to be incorporatea largenumber of lengthscales and parameters. These scales usedby studentsoutside of classtime that will addressthis need. Among are the resultof local fluid interactionswith either boundarysurfaces or ideasbeing considered are computer-graphichomework assignments, acrossdiscontinuities in the physicalproperties of the fluid.The art of specialproject-type lab setups, video-cassette instructional aids, and so- developinga correctdescription of anyphysical situation is in discerning nat games.Cunent materials and experience will bedescribed. the relativeimportance of eachof the parametersof the solution.One mightconsider solutions to a problemas a restrictedrange of valuesof theseparameters. In sucha case,these solutions would be locally valid in the parameterspace and are an asymptoticrepresentation of the global solution.The integrationof locallyvalid solutions into a globaldescrip- 11:15 tion is the essenceof singularperturbation methods. In this paper,the WW8. Farfieldcriteria for vibratingstructures: Numerical simulation on methodby whichcomputer algebra can be used to solveproblems involv- inga singularperturbation analysis in physicalacoustics will bediscussed. platesand cylinders.Christophe Journeau (Consulting Engineer for Centred'Etudes et de Recherchespour ia Discrttiondes Navires, DeAN Toulon, 83800 Tolon Naval, France), Thierry Robart, and FranqoisPellegrini (Centre d'Etudeset de Recherchespour la DiscretionAcoustique des Nayires, DeAN Toulon,83800 Toulon Naval, 11:00 France)
WW7. Creationof experiment-orientedmaterials to enhancea nonlab In orderto determinepractical criteria for thencarfield/farfield tran- acousticselective. S. A. Elder and M.S. Korman (PhysicsDepartment, sitionin the caseof underwatervibrating structures, the acousticfield U.S. Naval Academy,Annapolis, MD 21402} radiatedby platesand cylindersis simulated.Various mode shapes of
S107 J. Acoust.Soo. Am. Suppl. 1, VoL83, Spring1988 115thMeeting: Acoustical Society of America S107 thesestructures are computedfor differentfrequencies including their pc + 12%( 1dB), andpressure and velocity have a phasedifference ofless resonancefrequencies. These calculations are made for radiatorsof differ- than + 15ø are tabulated as a functionof frequency.The envelopeof these entsizes and are compared with theresults of a mechanicalsimulation for pointsis comparedwith the standard criterion (R•,2 andR•,L 2/26). a point-excitedribbed cylinder. Then the distances at whichthe pressure Finally,various criteria are proposed depending on whichproperty of the field is at + I dB of the farfieldvalue, the wave specificimpedance is farfieldis required.
FRIDAY MORNING, 20 MAY 1988 WEST BALLROOM A, 8:15 A.M. TO 12:00 NOON
SessionXX. Physical AcousticsIX: Cavitation, Bubbles,and Medical Acoustics
Ronald A. Roy, Chairman Departmentof MechanicalEngineering, Yale University,Box 2159, YaleStation, New Haven, Connecticut06520
ContributedPapers
8:15 8:4•
XX1. The Blake thresholdof a bubbleha•ng a radius-dependentsurface XX3. An activeacoustic backscattering technique for detectingtransient tension.Anthony A. Atchley (PhysicsDepartment Code 61 Ay, Naval cavitation.Ronald A. Roy, SameerMadanshetty, and Robert E. Apfel PostgraduateSchool, Monterey, CA 93943) (Department of MechanicalEngineering, Yale University, Box 2159 Yale Station, New Haven, CT 06520) The Blake thresholdis incorporatedin most theoreticalmodels of transient acoustic cavitation. In standard derivations, this threshold is Relativelydirty wateris insonifiedwith shortpulses (order 10/as)of calculatedfor a bubblehaving constant surface tension. However, such a high-frequency(= 1 MHz), focusedultrasound. The subsequenttran- bubbleis unstableagainst dissolution and not a realisticmodel for a cavi- sient cavitationbubble activity is detectedby a focused30-MHz pulse- tation nucleus.In this work, the Blake threshold is calculated for a more echodetector conlocally positioned relative to the insonifyingtransducer. realisticnucleus---a bubble having a radius-dependentsurface tension. A The eavitationdetector is sensitive enough to observesubmicron size bub- relationshipis derived for thisdependence, based upon the properties of a bles.Transient cavitation thresholds measured using this active detection suffactant-stabilizedcavitation nucleus. Two cases,appropriate for systemare as much as 40% lowerthan those measured using other tech- chargedand polar surfactants, are examined. The Blakethreshold for the niques(such as passive acoustic detection ). Thresholdsare measured as a formercase is identical to that of thestandard ease by approximately2%- functionof variousacoustic parameters (e.g., pulse length, pulse repeti- 20% for radii rangingfrom 3.0 pm down to 0.01/am. Therefore,it is tion frequency,etc.) andthe resultsare compared with previousinvesti- concluded that both unstabilized and surfactant-stabilized bubbles be- gationsas well as theoretical predictions. [Work supportedby NIH Grant comemechanically unstable under essentiallythe sameacoustic condi- No. I R01-CA-39374.] tions.Thus the useof the simpler,unrealistic nucleus in theoreticalmod- els resultsin no seriouserror. [ Work supportedby NPS Foundation.]
9:00
8:30 XX4. Brewsterangle light scatteringfrom bubbles in water:Observations and potential applicationsto the acousticsof natural microbubbles.S. XX2. Measurementof acousticalcavitation in liquid4He. JoelA. Nissen, M. B/iumerand P. L. Marston (Departmentof Physics,Washington E. Bodegore,L. C. Brodie,and J. S. Semura(Department of Physics, StateUniversity, Pullman, WA 99164-2814) Portland State University, Portland, OR 97207) Considerthe reflectionof light froma cleangas bubble in watersuch A piezoelectrichemispherical transducer was used to focushigh-in- that the light is polarizedwith its electricfield parallel to the scattering tensityultrasound into a small volumeof liquid helium. The transducer plane. The Fresnel reflection coefficientshould vanish when the local wasgated at its resonantfrequency of 566kHz, with gatewidths less than angleof incidencei is at Brewster'sangle ia = arctan(n- ' ) = 36.9deg, I ms, in order to minimize the effectsof transducerheating and acoustic wheren = 1.33is the refractiveindex of water.When thescattering angle streaming.The onsetof cavitationwas detected with small-anglescatter- 0 is close to that of the Brewster scattering angle, 0B ing of laserlight from the cavitationzone by microscopicbubbles. In the = ( 180 -- 2lB) m 106.2deg, this Brewstereffect should be stronglymani- superfluid,the pressureamplitude at thefocus was calculated based upon festedin the scatteringpattern [D. L. Kingsburyand P. L. Marston, the acousticpower radiated into the liquid, the geometryof the trans- Appl. Opt. 20, 2348-2350 ( 198i ) ]. In the presentresearch the firstlabo- ducer,and the nonlinearabsorption of sound.Above the lambdapoint, ratory observationsdescribed are of Brewstereffects in the scattering thismethod is complicatedby the scatteringof soundby bubblespresent patternsof bubblesrising through water having radii = 70/am.There is a in the liquid and on the surfaceof the transducer,and so a calibration noticeablereduction in thevisibility of fringesin thescattering pattern for curvefor the diffractedlight intensityat differentpressure amplitudes and scatteringangles = 0•. (The fringesarise from the interferenceof a far- temperatureswas obtained in thesuperfluid. The diffractedlight intensity sideray with the reflectedray. ) Microbubblesof naturalorigin can be was then usedto determinethe acousticpressure and extendthe cavita- importantin oceanand bioacoustics and Brewster effects may be useful in tion measurementsto the normal fluid. [Work supportedby NSF.] the opticalcharacterization of suchbubbles including the discrimination
S108 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spdng1988 115th Meeting:Acoustical Society of Amedca S108 of opticalscattering patterns of bubblesfrom those of particlesor cells. The scanninglaser acoustic microscope (SLAM) at 100 MHz and Brewsterangle scattering may also be usefulfor characterizationof the backscatteringacoustic technique (BAT) at 10-40 MHz wereused to surfactantcoating that is thoughtto occuron somemicrobubbles in na- examinecanine skin. Specimens from four animalsand from four loca- ture. [Work supportedby ONR.] tionson the animalwere analyzed biochemically and morphologicallyas well as acoustically.The meancoilagen and water concentrationswere 20% + 2% and 60% _+3%. respectively.An analysisof varianceof the SLAM dataand the biochemical data showed significant animal to animal differences and some differences due to location on the animal. At 100 9:15 MHz themean ultrasonic speed obtained with the SLAM was1632 4-_ 34 XXS. Enhancednonlinearity in a bubbly liquid consideredas a mixture. m/s and the mean attenuationcoefficient was 57 + 10 dB/mm. Using Erich Carr Everbachand Robert E. Apfel (Yale University,2159 Yale BAT the meau integratedattenuation coefficient was I I + 3 riB/ram at Station, New Haven, CT 06520) 25 MHz (midfrequencyrange). While the speedvalues fall within the rangeof valuespreviously reported for skin,the values for the attenualion During the pastdecade there has been an increasinginterest in the coefficientusing either SLAM or BAT are considerablyhigher than nonlinearbehavior of gashahhies in liquids.Recent papers by Sovietin- wouldbe predicted from literature values at 1-10 MHz. Thus theattenu- vestigators[e.g., Kustov et ul., SOy.Phys. Acoust. 32(6), 500-503 ationcoefficient is a strongerfunction of frequencythat the dataat lower (1986) ] have concentratedon phenomenaobserved when soundinter- frequencieswould suggest. [Work supportedby NIH andRR&D of the actswith a bubblylayer. These phenomena can includeacoustic solirons, Veterans Administration.] parametricradiation, self-focusing, phase conjugation, and othereffects thatoccur because the system is highlynonlinear. An investigationof the acousticnonlinear parameter of mixtureswas conducted to helpinfer the volumefractions of the mixturecomponents. When a bohhlymedium of knownair volumefraction is considered as a two-componentmixture, its 10:00 effectivenonlinear parameter may be calculatedusing mixture method- XXg, Off-resonance enntn'butions to acoustic hackscattering from ologiesJR. Apfel,J. Acoust.Sec. Am. 74, 1866-1869( 1983) ]. The mix- bubbleswarms, Kerry W, Commanderand Elan Moritz(Naval Coastal ture analysisin this work, whichis independentof any assumptionsre- SystemsCenter, Physical Acoustics Branch, Code 4120, Panama City, FL gatringbubble oscillation, yields nonlinear parameter values comparable 324O7) to thoseof the Sovietinvestigators. For certainvolume fractions, values of the nonlinearparameter of a bubblyliquid may be hundreds of timesthat Recentdiscrepancies between measurements of bubble populations of eitherthe gasor the liquid host.Both experimentaland theoretical nearthe surface in theocean by optical and acoustical methods have been work will be presentedand the implicationsof thisenhanced nonlinearity discussedin detailby Maclntyre[F, Maclntyre,"On reconcilingoptical will be discussed.[Work supportedby NIH throughGrant 5-R01- andacoustical bubble spectra in themixed layer," in OceanicW•itecap$, GM30419.] editedby E. C. Monohanand G. MacNiocaill(Reidel, New York, 1986), pp.75-94.] Another possible explanation for the very large number of smallbubbles found by acoustic methods is the potential for overestimates of bubblenumbers through the useof resonancescattering theory in the bubblepopulation calculations. In thiswork it is shownthat for some 9'.3O bubbledistribution of interest,the acoustic backscattering for a particular XX6. Measurementtincertainty assessment of thescanning laser genustie frequencyisactually dominated byscattering contributions from off-reso- microscopeand application to canine skin and wound. William nancebubbles. This effect is moreprevalent for highfrequencies, causing D. O'Brien,Jr., DianneL. Steiger(Bioacoustics Research Laboratory, overestimationof the numberof smallbubbles, which is preciselywhere Departmentof Electricaland ComputerEngineering, University of thetwo methods disagree. The more complete acoustic calculations may Illinois, Urbana, IL 61801), John E. Olerud, Mary Ann Riederer- notcompletely reconcile the bubble number estimates obtained from the Henderson,and George F. Odland{DepartmentofMedicine, University opticaland acoustical bubble spectra; however, they bring them closer of Washington,Seattle, WA 98195} together.[Work supported by ONT.[ The assessmentof measurementuncertainty of the scanninglaser acousticmicroscope (SLAM) hasnot been thoroughly evaluated experi- mentally.It is essentialthat the uncertaintyof any measurementbe fully understood.Both accuracy and precision were experimentally evaluated for the SLAM by measurementson homogeneousliquids of knownultra- 10:lS sonicproperties. Using aqueous solutions of bovineserum albumin, the attenuationcoefficient accuracy and precisionare + 12% and + 15%, XX9. Ultrasonicpropagation properties of artieularcartilage at 100 respectively.And fromDew Coming710, a siliconoil, the speed accuracy MHz. D. H. Agemura,W. D. O'Brien,Jr. (BioacousticsResearch andprecision are + 2.9% and + 0.4%, respectively.Further, an appli- Laboratory,Department of Electricaland ComputerEngineering, cationof the assessmentof precisionwas conducted using duplicate sam- Universityof Illinois,Urbana, IL 61801), J. E. Olerud,L. E. Chun,and piesof dog skin and wound tissue. From these evaluations of a heterogen- D.R. Eyre (Universityof Washington,Seattle, WA 98108) eoustissue, the estimated precision in themeasurement of theattenuation A pilotstudy on articular cartilage assessed the contribution of indi- coefficientand speed was + 16%and + 1.7%,respectively. [Work sup- vidualmatrix components to ultrasound propagation. The influence of portedby NIH Grant AM 21557and the RehabilitationResearch and collagenfibril orientation and collagen crosslinking was also assessed. DevelopmentService of theVeterans' Administration.] Sectionsof adultbovine articular cartilage taken both parallel and perpen- dicular(cross .sections) to the artieularsurface were examined using the scanninglaser acoustic microscope (SLAM) operatingat anultrasonic frequencyof 100MHz. A setof sampleswas evaluated that hadbeen 9:4• sequentiallytreated by enzymesto ( 1) removc85% of thechondroitin sulfate;(2) removeremaining GAGs, glycoproteins, and other noncolla- XX7. Propagationof ultrasoundin skin. Mary Ann Riederer-Henderson genproteins, leaving only the collagen fibril network; and (3) disruptthe (1057 Summit Avenue East, Seattle, WA 98102), John E. O!erud, collagenintermolecular crosslinks. Two striking observations were made: George F. Ocliand (Department of Dermatology, University of a profoundeffect of the"preferred" collagen fibril orientation on acoustic Washington,Seattle, WA 98195), Fred K. Forster (Departmentof wavespeed and a markedincrease in attenuationeocfficient when inter- MechanicalEngineering, University of Washington,Seattle, WA 98195), molecularcrosslinks were broken in the collagen.[This work was sup- DianneL. Steiger,and William D. O'Brien,Jr. (BioacousticsResearch portedby NIH GrantsAM21557, CA36029, and AR36794 and Rehabili- Laboratory,Department of Electricaland ComputerEngineering, tation Research and Development Services of the Veterans Universityof Illinois,Urbana, IL 61801) Administration.]
S109 J. Acoust.Sec. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America $109 10'.30 angle.The time domain correlation technique has been verified in a blood flow phantomsystem using both a bloodmimicking substance and por- XXI0. Ultrasonic propagationproperties of canineaorta at 100 MHz. cineblood. The transducermeasurement angle has been determined with- Felice Chu (BioacoustiesResearch Lab., Departmentof Electricaland in 5%, and pulsatileflow hasbeen measured up to 150beats/mix with ComputerEngineering, University of Illinois, Urbana IL 61801), Mary accuraciesbetter than 18%. Currently,this methodis beingexperimen- Ann Riederer-Henderson,J. E. O!erud (University of Washington, tally investigatedunder in vivoconditions, with the eventualapplication Seattle, WA 98108), and William D. O'Brien, Jr. (Bioacoustics beingthe diagnosis of venousthrombosis in humans.[Work supportedby Research Laboratory, Department of Electrical and Computer NSF/NHLBI Grant 1 ROI HL 37904.] Engineering,University of Illinois,Urbana, IL 61801) A pilotstudy examining canine aorta investigated the ultrasonic prop- agationproperties as a functionof locationand orientation along the aor- ta. Longitudinal,circumferential, intimal, and adventitial sections of ca- nineaorta were taken at sevenlocations along the aortafrom the anrtie 11:15 arch to the iliac bifurcation.Aortic mediaconsists of eircumferentially orientedeiastin fibers interspersed with a randomlyoriented network of XXI3. Ultrasonic attenuation coefficient assessment in rat liver from fibrils. Over the distancefrom the heart to the abdomen,the elastincon- carbon tetrachloride.David J. Boatright, Jr., W. D. O'Brien, Jr. centrationdecreases as the collagen concentration increases. The speedis (Department of Electrical and Computer Engineering), and J. shownto bestatistically significant with respectto locationand correlates W. Erdman, Jr. (Departmentof Food Science,University of Illinois, stronglywith orientation.The 28 speedvalues ranged from 1610-1859m/ Urbana, IL 61801) s with the averagebeing 1662m/s. On the other hand, the attenuation The ultrasonic attenuation coefficient is examined in rat liver due to coefficientis not shownto bestatistically significant with respectto orien- the effectsof inhalationof carbontetrachloride using the insertion-loss tationbut doescorrelate strongly with respectto location.The 27 attenu- technique@ 1.38:5,4.210, 7.015, and 9.820 MHz; it is theorizedthat rat ation coefficientvalues ranged from 48.5-131.8 dB/mm with an average liverswill developan increaseof collagen.Tissue constitutes of glycogen, of 91.2 dB/mm. [This work was supportedby NIH (}rants AM21557, protein,and water concentration arc assessed. The concentrationof these CA36029,and AR36794 and RehabilitationResearch and Development tissueconstitutes over an 1l-week period on the carbontetrachloride Service•of the VeteransAdministration. ] comparedto theattenuation coefficients at the four frequenciessuggest ultrasonicproperties can be modeledas a functionof constituteconcen- trations.At 1.385MHz, theattenuation coeffieient is statisticallysignifi- cantas a functionof glycogen,water, and protein.At 4.210 MHz, only proteinis statisticallysignificant; at 7.015MHz, bothglycogen and pro- 10:45 teinare statistically significant; and at 9.82MHz, noneof thetissue prop- ertiesis shownto besignificant with respectto theattenuation coefficient. XXll. Ultrasonic absorptionproperties in excessivelyfatty rat liver. [ Supportedby NIH Grant CA 36029.] Nadine B. Smith, W. D. O'Brien, Jr., and J. W. Erdman, Jr. (BioacousticsResearch Laboratory, Department of Electrical and ComputerEngineering, University of Illinois,Urbana, IL 61801) The ultrasonicabsorption coefficient was examined in rat liverdue to 11:•0 theeffects of 1% dietaryorotic acid measured using the transient thermo- XX14. Widebandacousticmlcroscopy oftissue. Christopher M. W. Daft electrictechnique @ IMHz. Dietaryorotic acid causes fat concentration (BioacousticsResearch Laboratory, Department of Electrical and to accumulateexcessively in ratsin a shortperiod of time. Tissueconsti- ComputerEngineering, University of Illinois, 1406W. Green,Urbana, IL tutesthat are particularlyimportant in absorptionare water, protein, collagen,and lipid levels. The concentration of thesetissues constitutes in 61801) and G. A.D. Briggs (Department of Metallurgy, Oxford rat liverover a (}-to10-day period on theorotic acid diet compared to the University,Parks Road, Oxford OXI 3PH, United Kingdom) absorptioncoefficient suggest that ultrasonicproperties can be modeled A scanningacoustic microscope (SAM) hasbeen used to measurethe as a functionof constituteconcentrations. Using the statisticalanalysis elasticproperties of tissuewith a resolutionof around8 pm (C. M. W. system,comparisons were made between the tissue constitutes and ultra- Daft et aL, Proc. 1986IEEE UltrasortiesSymposium, 945--948). This is sonicabsorption coefficient where comparisonsarc significantat the achievedby broadbandexcitation of the acousticlens, and the recording p < 0.05level. Lipid levels of thefatty liver were shown to bestatistically ofan undemodulated returning signal. This has the effect of separating the significanttothe absorption coefficient. Strong correlations ofboth colla- variouscontributions normally present in a SAM imageof tissue.A meth- genand protein with absorption, which havep values of 0.098and 0.071, od of analyzingthis information to yieldthe soundvelocity, acoustic im- respectively,were observed. [Work supported by NIH (}rantCA36029. ] pedance,section thickness, and acousticattenuation is described.Results are presentedfrom a sampleof skin tissue,and sourcesof experimental error are discussed.The elasticproperties of tissueappear to vary more substantiallyon thisscale than in macroscopicinvestigations. High values of aconsticvelocity and attenuationare observedin areaswith high con- 11:oo centrationsof structuralprotein and low water content. [ Work supported by SERC and GEC Hirst ResearchCentre. ] XX12. Accuratemeasurement of bloodflow usingultrasound and time domain correlation. Ilmar A. Hein and William D. O'Brien, Jr. (Department of Electrical and Computer Engineering,University of Illinois, Urbana, IL 61801 )
Usingtime domain correlation, the volume fluid flow in a vesselcan be accuratelyand precisely estimated without any previous knowledge of the 11:45 vesselsize, flow velocityprofile, or transducermeasurement angle. This XXl$. Acousticalquantification of thermal injury. Anne W. Holmes techniqueobserves the changein arrivaltime (insteadof changein fre- (Center for Bioengineering,University of Washington,Seattle, WA quency)of ultrasoundreflected from movingscatterers. Ifs vesselwith 98195), Fred K. Forster (Departmentof MechanicalEngineering, movingfluid is insonated by two ultrasonicpulses separated by time T, the University of Washington, Seattle, WA 98195), John E. Olerod scatterers within the ultrasonic beam will move some distance D between (Departmentof Dermatology,University of Washington,Seattle, WA the firstand secondpulse. There will bea difference•- in the arrivaltimes 98195), and Sam R. Sharar (Department of Surgery, University of of the two pulsessince the scatterershave moved.The scatterervelocity Washington,Seattle, WA 98195) canthen be calculated from the timedifference r. ff thevelocity is sampled at differentpositions within the vessel,the volumerate of flow can be In thisstudy, acoustic imaging was augmented by tissueeharaeteriza- calculated,along with thevessel dimensions and transducer measurement tion to moreaccurately quantify severity of thermalinjury. The acoustic
$110 J. Acoust.So•. Am.Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S110 attenuationcoeffcient in normaland burnedskin was investigated over creasedapproximately 30% to 5.7 Np/cm for burns.The differencesbe- the rangeof 10-30 MHz. Fourteennormal and 21 full-thicknessscald tweennormal and burnedskin were also characterized by the frequency burns( 1 and 2 rain) fromgoats were analyzed. The resultsof thisstudy dependenceofthe attenuation with a powerlaw of the form, af b. Charac- showthe importanceof choosingsites for attenuationestimates from the terizedin this manner,the fitting parametersexhibited large changes imagesince both normaland burnedskin cannot be assumedhomogen- when comparingvalues for normal and burnedskin. The parametera eous.From regionschosen in the reticulardermis, it wasfound that the increasedby more than a factor of 5 and the exponentb decreasedby attenuationcoefficient was significantly less for the burns.After averag- approximatelya factor of 2. Thesequantitative measures appear to be ingthe results from each group, the integrated attenuation coefficient over valuablecomponents in an overall acousticsystem for burn severity the frequencyrange studied was 8.3 Np/cm for normalskin, and de- evaluation.[Work supportedby NIH.]
FRIDAY MORNING, 20 MAY 1988 GRAND BALLROOM C, 8:15 A.M. TO 12:02 P.M.
SessionYY. SpeechCommunication VIII: SpeechProduction
JamesH. Abbs, Chairman WaismanCenter, Universityof Wisconsin,Madison, Wisconsin53706
Chairman's Introduetion---8:15
ContributedPapers
8:17 Hirano andhis colleagues [ Hirano et al. ( 1981) ]. The modelexhibits the relativemotions of the laryngealcartilages and the vibrationof the vocal YYI. Videostroboscopicimages associated with giottographic folds. The vocal folds of the model vibrate in a manner similar to observed waveforms.Gerald S. Berke, Bruce R. Gcrratt, and David G. Hanson vibrationsof humanvocal folds. The fundamental frequency of vibration, (VA MedicalCenter West Los Angeles, W 112C,Los Angeles, CA 90073 or pitch, increasesas tensionis increasedon the vocalfolds. Measure- andUCLA Schoolof Medicine,Los Angeles, CA 90024) mentsof thefundamental frequency of vibration,the pressures above and Knowledgeof vocalfold vibratory morphology in normaland patho- belowthe glottis,and the airflowthrough the glottisfor variouscondi- logicphonation is importantin understandingglottal airflow and acous- tionsof tension.mass, stiffness, glottal area, and profileof the vocalfolds tics.Although noninvasive measures of pronationsuch as photoglot- of the mechanicalmodel will be reported.The behaviorof the modelis tographyand electroglottography provide some information on vocal fold comparedto measurementsof pressure and airflow in previouslyreported movement,inferences are required regarding which glottal events corre- staticmodels and interpretedin light of currenttheories of vocal-fold spondto particularpoints on the giottographic signals. Thus confirma- vibration. tionof therelationship of glottographyto thevibratory movements of the vocalfolds is desirableand especiallyimportant when studying dys- phonia.Here a newtechnique is reportedin whichsequentially obtained 8:47 videostroboscopicimages are associated with specificvocal fold vibratory eventsrecorded giottographically. The systeminvolves digitizing giotto- YY3. An esophageaielectrode for studying posterior cricoarytenoid graphicsignals simultaneous with the horizontal sweep synch of the video muscle activity during speech.Mihoko Fujita, Christy L. Ludlow, signaland a 5-mssynchronization impulse that is alsorecorded on the Gayle E. Woodson,and Ralph F. Naunton (SpeechPathology Unit, audiochannel of a videorecorder. The advantages of thesystem include Building 10, Room 5N226, NINCDS, NIH, Bethesda,MD 20892 and theability to samplemany images and applicability to clinicalinvestiga- UCSD Schoolof Medicine, San Diego, CA 92103) tion. Validationof the techniquealong with the demonstrationof the imagesand waveforms obtained will bepresented. [Work supported by a An esophagealsurface electrode was developed for thestudy of poste- VA TechnicalMerit ReviewGrant. ] rior cricoarytenoic(PCA) muscleactivity during speechbased on an initial designof Woodsonand her colleagues.The electrodeconsists of two multiplestrand wires contained in a 5 Frenchpediatric feeding tube. It ispassed through one nasal passage into the upper part of theesophagus and picksup PCA activity throughthe anteriorwall of the esophagus. 8:32 Deepinhalation is a verifyinggesture with an absenceof activationduring sustainedmodal vowel production. Use was attempted with 19subjects; 3 YY2. Behaviorof a dynamicmechanical model of a larynx. Kevin Brown normalspeakers and 16 patients.The expectedactivation patterns were (Department of Mechanical Engineering,Massachusetts Institute of obtainedin 14 (73.7%) subjects.Problems that interferedwith obtaining Technology,Cambridge, MA 02139) and Cotinc Bickley (Room 36- accuraterecordings included a lack of descentinto the esophagusand 521, Department of Electrical Engineeringand Computer Scienceand pickupof phonatoryvibration. This electrodeprovides a valuabletool for Research, Laboratory of Electronics, Massachusetts Institute of phoneticand phonatorystudies of PCA activationpatterns in normal Technology,Cambridge, MA 02139) speakersand patientswith voicedisorders. A dynamicmechanical model of an adultlarynx was constructed. The componentsthat represent the cartilages were machined out of metal. The sizesof thecartilages were taken from measurements reported in theliter- 9:02 ature [Mane (1970); Kahane(1982)]. A mechanismwas designed to allow the arytenoidcartilages to be movedin a mannersimilar to the YY4. The crico-thyroidmuscle in voicingcontrol. Anders Lofqvist, complex,natural movements. The vocalfolds were simulated using natu- Thomas Baer, Nancy S. MeGart, and Robin Seider Story (Haskins ral rubber.The lengthof the vocalfolds was based on measurementsby Laboratories,2'70 Crown Street,New Haven, CT 06511-6695)
S111 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S1111 Initiation and maintenanceof vibrationsof the vocal folds require 9:47 suitableconditions of adduction,longitudinal tension, and transglottal X•7. Wisconsin x-ray microbeam system: Subject variables and airflow. This studyexplores the controlof voicingand voicelessnessin experimental procedures. Roxanne DePaul, Robert Nadler, speechwith particularreference to changesin the longitudinaltension of thevocal folds. Electromyographic recordings were made from the crico- Toyota Ishii, and Carl Johnson (Waisman Center, University of Wisconsin,Madison, WI 53706) thyroidmuscle in two speakersof AmericanEnglish and onespeaker of Dutch. The linguisticmaterial consisted of reiterantspeech made up of In this firstyear of operationof the x-ray microbeamlaboratory for CV syllableswhere the consonantswere voicedand voicelessstops, frica- the studyof speechmovements, certain procedures and methodshave tives, and affricates.The results indicate that the crico-thyroid activity beendevised and/or haveevolved from practical exprience. These include increasessubstantially at the transitionfrom a vowelto a voicelessconso- guidelinesfor therecruitment, screening, and selection of humansubjects, nantbut not during the transition from a vowelto a voicedconsonant. The alongwith acquisitionof informationregarding their anthropometric, activitylevel of the crico-thyroidis thushigher for a voicelessthan for a linguistic,neurological, educational, dental, and auditory characteristics. voicedconsonant. Measurements of the fundamentalfrequency at the Likewise,attempts have been made to deviseconsistent methods for beginningof a vowelfollowing the consonantshow the commonlyob- placementof articulatorypellets as well as for across-subjectstandardiza- servedpattern of higherF0 aftervoiceless consonants. It is thussuggested tion of thosepellet placements. In addition,for pelletson eacharticula- that the crico-thyroidmuscle shows a differentpattern of activationfor tory site,movement bandwidths and associated sampling rates have been voicedand voicelessconsonants with higheractivity for the latter. This empiricallydetermined. In parallel,methods have been tested and refined findingmost likely reflectsa differencein the tensionof the vocalfolds for simultaneouslysampling speech acoustic and other physiological sig- associatedwith the voicingcondition. The samemechanism also seems to nalsas well astheir postacquisitiontemporal alignment with pelletmove- contributeto the well-knowndifference between F0 at the beginningof mentdata. Other methods of experimentalimportance include techniques vowel followingvoiced and voicelessconsonants. for cuingsubjects and for presentationof utterancematerial. This presen- tationwill coverthese matters with thegoal of providingsociety members with additional x-ray microbeamsystem familiarity as well as a forum for identificationand discussionof yet other methodologicalissues. [Sup- 9:17 portedby NIH Grant NS-16373-8.] YY$. Fluid mechanicsof the velopharyngealtract. Ronald C. Scherer (DenverCenter for the PerformingArts, 1245Champa Street, Denver, CO 80204) 10:02 Predictionof geometricalcharacteristics of the velopharyngealtract (VPT) duringspeech would aid in determiningspeech adequacy and Y¾8. Wisconsinx-ray microbeamsystem: Acquired articulatory data interventioneffectiveness. Earlier work [beginningwith D. W. Warren and somepreliminary analyses. James H. Abbs, Robert D. Nadler, and A. B. DuBois,Cleft. Pal. J. 1, 52-71 (1964)] attemptedto predict CayleeNychis-Florence, and Nadine P. Connor (WaismanCenter, cross-sectionalarea of the VPT frommefisurements of mean flow through Universityof Wisconsin,Madison, WI 53706) the tract and mean pressuredrop acrossthe tract by usinga discharge To complementthe two previouspapers, it is usefulto providea more coefficient(DC) expressionwhere the DC is constant.In this study,the detaileddescription of data obtained on the new x-ray microbeam system. assumptionof a constantDC ischallenged through the use of a theoretical In the experimentto be described,two-dimensional movements were ac- equationapplicable to similarair channels.The equationtakes into ac- quiredfor pelletsplaced on themandible, tongue dotsum, tongue blade, countthe geometryof the VPT, and includeskinetic pressure terms for andtongue tip aswell as for two referencepellets on a verticalstrut at- the VPT entranceand exit and a viscouspressure term for the duct.The tachedto aneye glass frame (for headmovement corrections). Sampling predictedDC significantlyincreases with Reynoldsnumber (Re) (espe- ratesreflected individual movement speeds: 180 samples/s (SPS) for the cially below 1000), asymptotesto highervalues for largerVPT cross- tonguetip, 90 SPSfor thetongue blade, tongue dotsum, and mandible, sectionalarea, and is mostsensitive to geometrychange for low Re. The and 45 SPSfor the referencepellets. Four adult normalsubjects each predictedDC is higherthan foundin earliermodeling studies. Results produced50-75 repetitionsof the utterances"hot apple," "hat apple," suggestthat earlierphysical models were too simple and further empirical "haddock,"and "hassock." Primary analyses focuesd on thecontribution studyis justified. of thesefour movementsto anteriorlingual consonant productions, in- cluding2-D pelletpositions at palatalcontact, and the velocity,ampli- tude,and direction of thetrajectories in achievingthose positions. Posi- tionand trajectory patterns, along with variabilitydistributions, provide 9:32 certainperspectives on themotor control of thesegestures. [Work sup- portedby NINCDS Grant 16373-8.] YY6. Wisconsinx-ray microbeamsystem: Artieulatory pellet tracking specifications.Robert D. Nadler, Phillip E. Robl, Daniel J. Wahl, John R. Westbury (University of Wisconsin,Waisman Center, 1500 Highland Avenue, Madison, WI 53706-2280), and Osamu Fujimura 10:17 (AT&T Bell Labortories,Murray Hill, NJ 07974) YY9. Alterationsin lip and jaw kinematicsas a functionof speechrate. The x-raymicrobeam system for studyingtongue movements and oth- Nadine P. Connor, James H. Abbs, Caylee Nychis-FIorence er articulatorygestures has been in routineoperation for the pastyear. (University of Wisconsin,Waisman Center, Madison, WI 53706), and The precisionand accuracywith which the systemcan track pelletsis Michael P. Caligiuri (VA Medical Center, San Diego, CA 92161) dependentupon a numberof factors.These include the x-raysystem oper- ating parametersand characteristics,pellet size,background conditions, Manipulation of speechrate providesa meansfor examining varia- and imageprocessing algorithms used for real-timetracking. The results tionsin articulatorykinematics with natural alterationsin speechmotor of studiesthat addressthese issues, and their importancewith respectto timing.Such rate studieshave been popular, based upon expected insights interpretationand useof articulatorymovement data, will be discussed. into underlyingspeech motor control processes.However, resultsfrom Othermethodological issues of importanceto potentialusers include head prior work on kinematicvariations with speechrate havebeen equivocal. movementcorrection via theuse of referencepellets and empirical specifi- The divergentfindings may be due to taskdifferences, the few subjects cationof anatomicalreference planes for expressingthe acquiredspeech studiedin eachexperiment, and/or limitationsin the numberof taskrep- movementdata. The x-ray dosageincurred by subjectsstudied to datehas etitions. In this context,each of ten normal subjectsproduced syllable alsobeen carefully examined. Even whenusing peak spatial dose, a con- repetitionsequences and sentences with relativelylarge numbers of repeti- servativeestimate, subjects can participate in repeatedspeech production tionsfor each.Multiple measuresof lip andjaw kinematicswere applied, studieswithout exceedingthe yearly Federalwhole body doselimit. The includingsome previously shown to be particularlyrevealing in studiesof statusof the systemand planned updates will alsobe presented. [ Support locomotion,arm, and hand movements.Findings indicatedpreviously for this facilityis providedby NINCDS (NS- 16373).] unreportedspeech rate-related variations in peakvelocity-amplitude ra-
Sl 12 J. Acoust.$oc. Am. Suppl.1, Vol. 83, Spdng1988 115th Meeting:Acoustical Society of America $112 tios,syllable-to-syllable magnitude and timing, vowel-consonant versus latorsmore rapidly when closing into voicelessversus voiced obstruents, consonant-vowelmovement durations, and peak velocity time-shifts. The whichmay contributeto shortervowels preceding voiceless obstruents. resultsprovide some new perspectiveson speechkinematic control. Despiteseveral acoustic studies of thisgeneral phenomenon in English- [Work supportedby NINCDS Grant 13274-12.] learningchildren, however, little developmentalinformation is available concerningarticulato• characteristics of children's post-vocalic conso- nants.Although children show acoustic vowel duration differences in thesecontexts, it is not knownwhether articulatory differences occur in their productions,as they do for adults.One purposeof thisstudy was to 10:32 obtainkinematic data from childrenconcerning this issue.Articulatory peakvelocity data for upperlip, lowerlip, andjaw movementswere ob- YY10. Labial closing kinematics reconsidered.Gerald D. Lame and tainedfor eightchildren in eachof threeage groups and eightadults. A RichardO. DeWaid (Departmentof Linguistics,University of Texasat peakvelocity difference for [p] vs [b] closinggestures was observed for Austin, Austin, TX 78712) eachof the fourage groups. There was also a tendencyfor the difference Ladefogedand Maddieson [UCLA Working Papers 64, 1-137 between[p] and lb] peakclosing velocity to increasewith age. (1986) ] havespoken of "the forcefulcoming together of the tongueor lips in a stopconsonant" as "an attemptto throwone part of the vocal tractthrough another" (p. 3), whileGracco and Abbs [Exp. BrainRes. 65, 156-166 (1986) ], in examiningclosure into the first/p/for "sapap- ple," foundpeak velocity was attained well beforecontact. This decelera-
tioninto closure, plus "relatively high correlations between the amplitude 11:17 and peakvelocity of individualmovements" were taken as evidence that "the labial closuregoal was not achievedsimply by slammingthe lips YY13. Native speakersof Spanishand English compensateequally well together"(p. 159). While thesedescriptions are not necessarilycontra- for a bite block in producingvowels. J. E. Flege (Biocommunication, dictory,they do suggesttwo differentpossible movement types. Gracco Universityof Alabamaat Birmingham,University Station, Birmingham, and Abbs'smooth, centrally peaked velocity profiles resemble those of AL 35294) pointingor reaching.Ladefoged and Maddieson describe, instead, a strik- inggesture. In orderto distinguishthese, special care must be taken ( 1) to A recentdetailed study [Flege etal., J. Acoust.Soc. Am. 83, 212-228 ascertaintime of contactaccurately, (2) not to distortevents near impact (1988) ] providedauditory and acousticevidence that talkersdo not com- by excessivelow-pass filtering, and ( 3 ) to avoideffects of secondarylabial pensatecompletely or instantlywhen producing vowels with a biteblock. articulations. When these precautions are taken, closure into This study examinedcompensation physiologically. Adult native speak- /p/ resemblesa blow or tap, with minimal if any decelerationbefore ersof Englishand Spanishproduced nonsense disyllables (/bVba/) in a contact. Spanishor Englishcarrier phrase. Glossometry was used to measurethe verticaldistance of the tonguefrom the hard palateat four sensorloca- tionsin multipleproductions of thestressed vowels/i/,/e/,/a/,/o/, and /u/. Even thoughSpanish has fewer vowelcategories than English(5 vs 15), the Spanishand Englishsubjects showed equal variability in tongue positions(standard deviationsof 0.746 and 0.756 ram). The subjectsin 10:47 bothgroups showed significantly greater variability in thebite-block than YY11. Extracting dynamic parametersfrom articulatory movement. normal-speechcondition (means s.d.'s of 0.821 vs 0.722 ram). Therewere Richard S. McGowan, Caroline L. Smith, Catherine P. Browman, and cleardifferences in the averagetongue positions for Spanishand English BruceA. Kay (HaskinsLabs, 270 Crown Street, New Haven,CT 06511) vowelsspoken normally. The eightsubjects in bothgroups showed tongue positiondifferences (unsigned) of only about 1 mm when producing In orderto modelthe movementsof the speecharticulators using a vowelsnormally and with a bite block.The effectof conditionon tongue dynamicalsystems approach, it is necessaryto specify( 1) the specific positionswas not significant for either group. [ Work supportedby NIH. ] formof thedynamical equation, (2) theportion of thearticulatory trajec- tory beingdescribed/generated by the equation--the"window," and ( 3 ) the valuesof the coefficientsto be usedin the equation.In this work, a dampedmass-spring dynamical system is assumed,as usedin previous taskdynamic modeling [ Saltzmanet al., J. Acoust.Soe. Am. Suppl.1 82, S15 (1987)]. A programwas developedto determinethe valuesof the coefficientsfor the equations,using various definitions of thearticulatory 11:32 window.The programuses a nonlinearleast-squares algorithm to deter- mine the bestfit betweenthe observedtrajectories and the trajectories YYI4. Intonational and rhythmic correlates of stress clash. generatedusing a rangeof valuesfor the coefficients.This programis Mary Beckman,Kenneth dcJong(DcpartmcntofLinguistics, 204Cunz currentlybeing tested on simulateddata with knownparameters in order Hall, Ohio State University, Columbus,OH 43210), and Jan Edwards to establishthe accuracyand limitations of the procedure.There will bea (Hunter Collegeof Health Sciences,CUNY, New York, NY 10021) reporton the reliabilityof the analysesperformed using this approach, Phonologistssupport a directly metrical representationof stress with an emphasison numericalconsiderations. [Work supportedby rhythmsby observingthat whenone stress "beat" follows closely on an- Grant BNS-8520709 from NSF and Grants HD-01994 and NS-13617 other, the rhythmic clashseems to be remediedeither by retractingthe from NIH. ] first stress,as in Chineseexports, or by increasingthe interstressinterval, sothat thepop of pop posed is longerthan that ofpoppa posed. Cooper and Eady [J. Mcm. Lang.25, 369-384 (1986) ] havequestioned the empirical validityof suchobservations by showingthat the firstsyllabic's duration is not longerin phraseswhere stress should have retracted to it; nor is it 11:02 longerin ChrisMack compared to ChrisMacKensie. Recent experiments, YY12. Kinematic characteristicsof children'spost-vocalic labial stop however,support the notion of rhythmicdash even if theprecise formula- consonants.Bruce L. Smith and Ann McLean-Muse(Department of tion of metricalphonologists is wrong.It wasfound ( 1) that syllables CommunicationSciences and Disorders, Northwestern University, perceivedas bearing a retractedstress have longer durations relative to the Evanston, IL 60208) lessstressed following syllable, (2) that phraseswith retractedstress have differentdistributions of pitchaccents in the intonationpattern, and (3) In English,vowels tend to belonger preceding voiced versus voiceless that althoughthe pop ofpopposed is notlonger than that ofpopopposed, it obstruents.Although perceptual factors are often thought to beprimarily is longerthan that ofpoppaposed. These results are usedto supportan responsiblefor thiseffect, part of the differencemay alsobe a resultof indirectmetrical representation in whichstress patterns are readfrom a inherentspeech production factors. Adults commonly move their articu- prosodicconstituent tree that includesintonational structure.
Sl 13 J. Acoust.Sec. Am. Suppl.1, Vol. 83, Spring1988 115th Meeting:Acoustical Society of America S113 11:47 motorcontrol instructions in theproduction of speechoutput. This work reportson thecollection of phonologicalspeech error data in a controlled YY15. Phonological constructs:Electromyographically obtained speech laboratory environment using Basmajian electrode EMG in various error evidence. Richard A. Mowrey (594 BessererStreet, Ottawa, speechmuscles, as well as audio recordings. Both integrated EMG signals Ontario KIN 6C9, Canada) and Ian R. A. MacKay (Phonetics and singlemotor unit analysisare presented.The resultslead to the con- Laboratory, LinguisticsDepartment, University of Ottawa, Ottawa, clusionthat published speech error corpora based on field-collected errors Ontario K1N 6N5, Canada) areskewed by thecombined problems of collectiontechnique, categorical perception,and hearer normalization, and that this skewing leads to cir- In recentyears, speech error data haveoften been cited as evidence for cularityin thearguments based on thesecorpora. This evidence provides the existenceof, or for the psychologicalreality of, certainphonological supportfor individualmotor instructions as phonologicalprimitives in constructs,such as the phonemeor thedistinctive feature. Further claims motorspeech control, and does not supporthigher-order motor control havebeen made concerning the role of theseunits in the processingof unitssuch as features and segments.
FRIDAY MORNING, 20 MAY 1988 GRAND BALLROOM B, 8:30 TO 11:35 A.M.
Sessionzz. PsychologicalAcoustics VI, PhysiologicalIV, andNoise VI: Effectof Noiseon the Auditory System
Julia D. Royster,Chairman EnvironmentalNoise Consultants, Inc., P.O. Box 144, Cary,North Carolina27512-0144
Chairman's Introduction---8:30
ContributedPapers
8:35 noisesrepresenting CRP configurationswith an unequalnumber of bladeson eachrotor. Included in eachexperiment were five conventional ZZ1. Objectivemeasures of theannoyance due to automobilepassenger turbopropand five conventional jet takeoffs.Each noise was presented at compartmentnoise. John S. Lamancusa(Mechanical Engineering threelevels to subjectswho judged the annoyance of eachstimulus. Anal- Department,Penn State University, University Park, PA 16802) ysesof thejudgments examine the effects on annoyanceof bladepassage The annoyanceattributable to complexnoises consisting of multiple frequency,tone-to-broadband noise ratio, and aircrafttype. The annoy- puretones (which may or maynot be harmonically related) and substan- anceprediction ability of currentnoise metrics is alsoexamined. tialbroadband noise is not easily quantifiable. It hasbeen shown that dBA is a poor indicatorof loudness[Hellman and Zwicker,J. Acoust.Soc.
Am. 82, 1700-1705(1987) ], and a numberof othermeasures have been 9:05 proposed.In thisstudy, the results of subjective preference tests of passen- ger compartmentnoise are comparedto objectivemeasures of loudness ZZ3. Quality assurancein automatedmonitoring of high-energy andannoyance. Eight synthesized variations of thenoise experienced at impulsivesounds. George A. Luz, William A. Russell, and Nelson full throttleacceleration were ranked by 40 subjectsaccording to their D. Lewis (U.S. Army EnvironmentalHygiene Agency, Aberdeen desirability.All sampleswere normalized to 101-dBlinear. These results ProvingGround, MD 21010-5425) are comparedto the objectiverating methods:AI (articulationindex), ISO 532B(Zwicker method), Stevens Mark VI I, CRP (CompositeRat- U.S.Army policy is to measurenoise levels whenever the projections ing of Preference),and PNL (perceivednoise level). The CRP offersthe ofnoise contouring computer programs show a noiseenvironment "unac- bestcorrelation with subjectiveresults. The resultsof Mark VII, and ISO ceptable"for residentialuse extending beyond the Federalinstallation's 532Bare markedly improved by theaddition of empiricallyderived cor- boundary.The threshold for "unacceptable" isan A-weighted day-night rections. level(DNL) of 75 dB for aircraftand a C-weightedDNL of 70 dB for high-energyimpulsive sounds. For adequatequality assurance (QA) in monitoringhigh-energy impulsive sounds, statistical decision criteria are employedincluding threshold, duration, number of eventsper unit time, peak-to-SELdifferences, and event probabilitydistributions. The 8:50 strengthsand weaknesses of eachQA criterionis discussed. ZZ2, Advancedturboprop aircraft flyover noise: Annoyance to counter- rotating-propellerconfigurations. David A. McCurdy(NASA Langley ResearchCenter, Mail Stop463, Hampton, VA 23665-5225) 9:20
Twoexperiments were conducted to quantifythe annoyance of people ZZ4. Annoyance factors for common neighborhoodnoise. Lynn to theflyover noise of advancedturboprop aircraft with counter-rotating S. Alvord (Departmentof CommunicationDisorders, 1201 Behavioral propellers(CRP). The objectiveswere: ( 1) determinethe effects of tonal ScienceBuilding, University of Utah, Salt Lake City, UT 84112) contenton annoyance;and (2) compareannoyance to advancedturbo- propaircraft with annoyanceto conventionalturboprop and jet aircraft. Neighborhoodnoise is treatedseparately from aircraft or trafficnoise A computersystem was used to synthesizerealistic, time-varying simula- in mostlocal noise ordinances. Annoyance factors for commonneighbor- tionsof advancedturboprop takeoff noise. For the first and second experi- hoodnoise were rated by 63 subjectsin Salt Lake City, Utah. Subjects ments,respectively, the systemgenerated 27 noisesrepresenting CRP wereasked torate/on ascale of 1to 5, the degree towhich each factor configurationswith an equal numberof bladeson each rotor, and 35 contributedto total annoyance.Most frequentlymentioned annoying
Sl 14 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring 1988 115thMeeting: Acoustical Society of America S114 noisesources were dogs (38.1%), followedby sirens(12.7%). The fol- 9:35 lowingsources were also identified at higherthan 3% incidence:garbage trucks,buses, children playing, doors slamming, noisy neighbors, helicop- ZZS. An overviewof occupationalnoise regulations in Canada.Deirdre ters, and metal stairways.Highest-rated annoyance factors were: loud- A. Benwell (Bureau of Radiation and Medical Devices, National Health ness,time of occurrence,frequency of occurrence,sound quality, and and Welfare, Room 233, EHC, Ottawa K1A 0L2, Canada) interferencewith sleep.Relative importance of annoyancefactors found in thisstudy differ from those of previousstudies, which dealt more with In Canada,there are occupational noise regulations at bothprovincial aircraftor trafficnoise. Fear and interference with conversation and sleep andfederal levels of government.An overviewof presentCanadian occu- areof lessimportance for "neighborhood"noise than aircraft and traffic pationalnoise legislation is given.In addition,recent Canadian activities noise[T. J.Schultz, J. Acoust. Soc. Am. 64, 377-405(1978) ]. For neigh- concerningoccupational noise exposure standards, guidelines, and other borhoodnoise "meaning" the soundportrays is an importantfactor in important backgrounddocuments are described.The variousmethods certainsituations. The importanceof "situationalfactors" found here usedto assesscompensation for occupationalhearing loss are alsosum- emphasizesthe needfor increaseduse of specificprohibitions in local marized. Some recommendations are made for future activities in this noise ordinances. area.
9:50-10:05
Break
10:05 10:35
ZZ6. The critical exposure in acoustic trauma. W. Dixon Ward ZZ8. Altered susceptibilityof the inner ear following repeated noise (Hearing Research Laboratory, University of Minnesota, 2630 exlmsure.Daniel J. Franklin, BardenB. Stagnet,Brenda L. Lonsbury- UniversityAvenue S.E., Minneapolis,MN 55414) Martin, and Glen K. Martin (Department of Otorhinolaryngologyand CommunicativeSciences, Baylot College of Medicine, Houston, TX As oneincreases the intensity Iofa noiseexposure of fixedduration t, 77030) the resultingdamage grows gradually up to a criticalpoint at whichthe damageincreases precipitously. The intensityat whichthis occurs, how- The susceptibilityof the innerear to repeatednoise exposure was in- ever,is notindependent of theduration, so it cannotbe called a "critical vestigatedin therabbit using tests of behavioralthresholds and distortion- intensity."Neither is this point dependent only on thetotal energy of the productemissions (DPEs). Behavioralthresholds, measured by a classi- exposureIt: Lessenergy is requiredat shorterdurations in orderto reach cal-conditioning technique, were collected at 11 frequencies thecritical point, so even "critical energy" would be a misnomer.Instead, representativeof the rabbithearing range. Acoustic distortion products at theimplication of an extendedseries of experimentsin whichchinchillas the2fl - f2 frequencywere measured as both DPE audiograms,genera- weregiven single exposures of a fewminutes to 1.5days at levelsup to 120 ted by equilevelprimaries at 45, 55, and 65 dB SPL, and input-output dB SPL is thatthe critical exposure appears to bedefined by 12t = C, or functionsin 5-dBsteps at ninefrequencies. Following acquisition of con- equivalently,p4t: k, wherek is about108 Pa 4 s for thechinchilla. This trol measuresof DPEs andbehavioral thresholds, DPEs wereanalyzed at valueis in agreementwith recentresults defining critical exposures to regular intervalsduring both exposureand recoveryperiods and com- impulsenoise in the chinchillareported by Pattersonand Hamernik [J. pared to correspondingbehavioral measures. In four animals, following Acoust.Soc. Am. 81, 1118-1129(1987) ]. The meagerdata in the litera- repeatedepisodes of exposure(octave band of noisecentered at I kHz at turesuggest that for man, k maybe around 10 "• Pa 4 s. [Worksupported 95 dB SPL) and recovery (3 weeks), both DPE amplitudesand be- by NINCDS Grant 12125.] havioralaudiograms revealed an increasingresistance to the effectsof noiseexposure. Comparison with the amountof amplitudereduction in a 4-kHz DP causedby a 5-rainexposure during the preexposure periods to a 95-dB SPL toneat 4.215 kHz ( 1/2 octavebelow the geometricmean of the primaries)showed a poor correlationwith the lessenedsusceptibility to repeatednoise exposure. The conceptdescribing a dynamicmechanism 10:20 of repairof sensoryelements proposed by a numberof investigatorsto explainrecovery from soundexposure may accountfor the increasing ZZ7. Influencesof temporalspacing of impulsenoise on temporaryand resistanceto the effectsof repeatednoise exposure as detectedby both permanentthreshold shifts. Richard W. Danielson (Cailler Center, DPEs and behavioralresponses. [Work supportedby NINCDS. ] Universityof Texas at Dallas, Dallas TX 75235), Donald Henderson, and SamuelS. Saunders(Department of CommunicationDisorders and Sciences,State University of New York at Buffalo,Buffalo, NY 14266)
Chinchillaswere exposedto impulsenoise conditions with equal acousticenergy and spectra but with threedifferent temporal patterns and 10:50 two differentamplitude levels ( 135and 150dB SPL, counterbalancedby ZZ9. Intracochlearpressures resulting from brief highintensity stimuli. 15-dB trade-offsin numberof events). Auditory-evokedresponses were James H. Patterson, Jr., Ben T. Mozo (U.S. Army Aeromedical usedto measureTTS and PTS. Degreeof hearingloss and patternof ResearchLaboratory, P.O. Box 577, Fort Rucker, AL 36362-5292), recoverywere not uniform among the groups,contrary to assumptions Karl Buck, and Laurent Decory (Franco-German Institute of relatedto theequal energy hypothesis. Relative to hearingloss after expo- Research, Saint-Louis, France) sureto oneimpulse/second: ( 1 ) lesohearing loss was seen after exposures to burstsof 1 impulse/50ms and (2) greaterhearing loss was seen when A previousstudy has shown similar hearing loss and cochlearinjury high-levelimpulses occurred in salvos(pairs of impulsespresented at l-s fromexposures to impulses differing in peakpressure, but having the same intervals, with 50-ms interpulse interval in each pair). Findings will be energy and Fourier pressurespectrum. Nonlinear processesin the middle comparedto reportsof a possible"period of vulnerability,"during which ear or transmissionthrough the externaland middleear may renderdif- hearingloss may be more extremeif the ear is exposedto additionalim- ferentstimuli more similar once they reach the cochlea, thus accounting pulsenoise while the ear is stillrecovering. [Work supportedby NIOSH for the equivalentinjury observed.In order to explorethis possibility, and the Departmentof the Army.] pressuresin thebasal turn of thecochleas of eightanesthetized chinchillas
$115 d. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115thMeeting: Acoustical Society of America Sl 15 were measuredfor impulsespresented at levelsbetween 88-and 147-dB theTC variables.The percentageof outerhair cell loss showed an orderly peak.This included the impulses used in theprevious injury study. While and systematicincrease as PTS increasedand as TC variableschanged nonlinearitieswere noted at the higherintensities, there was a remarkable acrossthe entirerange of testfrequencies. The innerhair cellswere essen- preservationof the pressure-time histories. In additionto theseimpulse tially unaffected.These results show that thereis a systematicchange in stimuli,intracochlear pressures were measured using brief tone pips the TC variablesthat definethe quality of tuningas hearing loss progres- whoseenvelope was Gaussian. Frequencies of 700, 1375,2000, 2700, and sivelyincreases and that these changes are clearly related to outerhair cell 4100Hz werechosen to span the frequency response of thespeaker. With losses.[Research supported by NIOSH andDOD. ] thesestimuli, nonlinearities appeared as both harmonic and subharmonic distortion as the intensities increased.
11:20 11:05 ZZll. "Enhanced" evoked responseamplitudes in the chinchilla ZZ10. Changesin frequencyselectivity in the chinchillafollowing a noise following acoustic trauma. R. Salvi, S. Saunders,N. Powers, and M. induced permanent threshold shift. Robert I. Davis, William A. Gratton (Department of CommunicationDisorders and Sciences, A. Ahroon,and RogerP. Hamernik(Auditory Research Laboratory, 109Park Hall, SUNY Universityof Buffalo,Buffalo, NY 14214) SUNY at Plattsburgh,Plattsburgh, NY 12901) The amplitudesof auditory-evokedresponses recorded from chronic Evoked-potentialtuning curves (TC) wereobtained on 106chinchil- electrodes in the inferior colliculus of the chinchilla were measured before las before and after acoustic overstimulation in order to assessthe effects andafter acoustic trauma. Acoustic trauma was induced using a 2-kHz of themagnitude of hearing loss on frequency selectivity. Pre- and postex- continuous tone that resulted in either 30-40 dB of TTS or PTS between posuremeasures of auditorythresholds and maskedthresholds (simulta- 2-8 kHz. The high-intensityexposures resulted in systematicchanges in neoustone-on-tone paradigm) wereobtained in eachanimal at 0.5, I, 2, 4, the input/outputfunctions of the evokedresponse. The moststriking 8, and 11.2kHz, usingthe evokedauditory response recorded from the changewas an increasein the maximumamplitude ("enhanced") of the inferiorcolliculus. Three TC variables(Q-10 dB, tail-tipdifference, and evokedresponse atfrequencies below and at thelow-frequency edge of the thehigh-frequency slope) and sensory cell losses were compared to the hearingloss (0.5 and2 kHz). Bycontrast, the maximum amplitude seen amountof noise-inducedpermanent threshold shift (PTS) producedby a at frequenciesnear the middle of thehearing loss or itshigh-frequency varietyof noiseexposures. Based upon large sample averages, frequencies border(4 and8 kHz) wasgenerally depressed. In addition to thechange showingPTS>20 dB alsoshowed statistically significant differences be- in maximumamplitude, there were also changes in theslope of the evoked tween pre- and postexposuremeasures of all three TC variables.For responseinput/output functions. The results will be related to thepattern 10< PTS< 20dB onlythe tail-tip difference showed a statisticallysignifi- of hair cell lossas well as to possibleunderlying neural mechanisms. cantchange, while for PTS<10 dB therewere no measurablechanges in [Work supportedby NIH R01 NS16761and NS23894.]
FRIDAY MORNING, 20 MAY 1988 WEST BALLROOM B, 8:30 A.M. TO 12:05 P.M.
SessionAAA. UnderwaterAcoustics VIII: UnderwaterPropagation
Kenneth E. Gilbert, Chairman TheNational Center for PhysicalAcoustics, University of Mississippi,University, Mississippi 38677
Chairman's Introduction---8:30
ContributedPapers
8:35 8:50
AAAI. PARQUES: A range-dependentenvironment propagation loss AAA2. Adaptationof a three-dimensionalray tracingcode for usewith model.Bruno Grandvuillemin(DCAN Toulon-GERDSM, Six-Fours griddedinput oceanographicmodels. J. Dougherty,A. E. Newhall,J. ies Plages83140, France) F. Lynch (Department of Ocean Engineering, Woods Hole OceanographicInstitution, Woods Hole, MA 02543), and C. S. Chiu (Institute for Naval Oceanography,NSTL, MS 39529) A proposalto solvethe Hclmholtz'sequation by meansof the parabol- icapproximation in a two-dimensionalrange-dependent environment will In orderto be ableto adaptthe Hamiltonianray tracingprogram begiven. It willmake clear the solution for a quadraticdevelopment ofthe (HARPO) developedby NOAA's Wave Propagation Laboratory for use propagationoperator that allows the consideration of a largerangular with griddednumerical oceanographic models, three different ap- apertureof the source. The numerical solution is given by an algorithm of proacheshave been employed. The firstmethod employs using a 3-D finitedifference using the Crank-Nicholson method. The problem of rec- Fourierdecomposition of the sound-speed field, giving one the ability to tili•ear, horizontal,and obliqueinterfaces between two fluid mediais usefast Fourier techniques and spatially filter input data but being limited solved.The problem of a rangevariable sound velocity profile is treated by to a constantgrid with mesh determined by thehighest spatial frequency conditionsof continuitybetween two media with the given profiles. The alonga givencoordinate. The secondtechnique employs an empirical regressivewave is alsocomputed. The resultsof numericalsimulations orthogonalfunction analysis of the sound-speedfield, the coefficients of executedon thearray processor FPS 164are presented.Also are shown whichare splined to ensurethe continuity requirements of the ray trace. comparisonsbetween theoretical computation and experimental data for Thismethod allows one to: ( 1) simplyfilter the field by mode energy: (2) range-independentand range-dependent environment. projecteasily on oceanographic modes; and (3) varythe sampling to meet
$116 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America $116 the localocean conditions, thus saving on computermemory require- In oceanacoustics, the introductionof rangediscontinuities, for ex- ments.The final techniqueemployed is "featurehanging," which com- ample,the water-to-icecanopy surface, creates a mixedboundary value binesusing analytic forms for oceanfeatures and gridded input for their problem.In thispaper, an exactsolution of certainmixed boundary value positions.This isan extensionof theexisting HARPO featureto beconsis- problemsis discussedusing the Wiener-Hopfmethod. A keyattribute of tentwith currentoceanographic models. Example calculations using each this approachis that it is not fundamentallynumerical in nature and techniquewill be presented. allowsadditional insight into the mathematicaland physical structure of theacoustic field due to rangediscontinuities. The problemdiscussed here is a canonicalone: A planewave is incidentupon a planarsurface where the boundarycondition changes from Dirichlet (free surface)to Neu- mann (hard). The boundaryconditions addressed in this problemare 9:05 highly idealizedrenditions of what happens,when, say, a plane wave is incidentupon a water-to-icecanopy surface; nevertheless, important fea- AAA3. The time-marchedFFP for modelingocean acousticpulse turesof the diffractionprocess are producedhere, and the solutiongives propagation.Michael B. Porter ($ACLANT UnderseaResearch Centre, considerableinsight into the process. The solutionof thediffracted poten- APO, NY 09019) tial is partitionedinto two components:a field consistingof cylindrical Traditionalocean acoustic models for cw (time-harmonic)signals wavesweighted by a polar gain function,resulting from the artificial may be extendedin an obviousfashion for broadbandproblems using sourcecreated by the discontinuityin the boundary;and a fieldcontaining Fouriersynthesis. Alternatively, many of thesemodels may alsobe solved a residuecontribution which restoresfield continuity along the line corre- directlyin thetime domain that raises the questions of whatis the optimal spondingto specularreflection. Contour plots of equalpressure ampli- solutionand under what conditions.A time-marchedfast-field program tudeshow how the component fields superimpose such that the boundary hasbeen developed in whichthe solutionis representedas a sumof Four- conditionsare maintained,and how energyis redistributedacross the ier componentsin range with time and depth-dependentamplitudes, angularspectrum in the diffractionprocess. The latter is relatedto mode a(z,t;k). Theseamplitudes each satisfy a simplehyperbolic equation that couplingdue to boundarychanges in waveguideproblems. [Work sup- isdiscretized in depthusing finite elements and in timeby a simpleexplicit portedby ONR. ] integrator.Snapshots of the oceanacoustic field are then obtainedin the usualfashion using an FF-'rto assemblethe variousspatial Fourier com- ponents.Examples of the methodapplied to pulsepropagation in certain oceanacoustic scenarios will be presentedwith particularregard to the questionof the strengthsand weaknessesof the time-marchedFFP versus 9:50 synthesisfrom time-harmonicFFP solutions. AAA6. Gaussianbeams and salt tables. H. P. Bucker (Code 541, Naval OceanSystems Center, San Diego, CA 92152)
In the studyof manyocean acoustic systems it is desirableto perform
9:20 veryrapid calculations of the soundfield. For mostproblems of practical interest,2«-D models provide accurate results. That is,the sound propa- AAA4, Intrinsic modesrevisited. F. Xiang, M. Cada (Departmentof gationis differentalong different radials but the interchangeof energy Electrical Engineering,Technical University of Nova Scotia, Halifax, betweenradials can be neglected.In this case,it is possibleto use salt Nova Scotia B3J 2X4, Canada), and L. B. Felsen (Department of tableswhere sound angle level and time is tabulated as a functionof range. ElectricalEngineering and ComputerScience, Polytechnic University, Thispaper examines the compatibility of theGaussian beam propagation Farmingdale,NY 11735) modeland salt tables.Special items covered are the partitionof beams betweentables, receiverson the surface and bottom boundaries,and the Intrinsic modes[J. M. Arnold and L. B. Felsen,J. Acoust.Soc. Am. effectof beamdisplacement associated with boundaryreflection. [Work 76, 850-860 (1984)] are wave objectsthat propagatewithout coupling partly supportedby the NORDA NOP Program.] upslopeor downslopein a homogeneousocean wedge above a penetrable bottom.They havebeen observed experimentally and havebeen referred to as wedgemodes in that context[F. B. lensenand C. T. Tindie, 1. Acoust.Soc. Am. 82, 211-216 (1987)]. The exact analytic form of the intrinsicmodes (IM) involvesa spectralintegral which, for smallslopes, 10:05 can be reducedasymptotically to a simplerintegral expression that leads to the conventionaladiabatic modes where these apply but is validalso AAA7. A space-timesplit-step algorithm for pulse propagationin throughthe upslopecutoff transition. Except for smallslopes, this simpli- refracting media. Johanan L. Codona (AT&T Bell Laboratories, fied form upslopeor downslopeis flawedby energybalance defects be- Whippany,NJ 07981) yondupslope cutoff and by inaccuraciesof the field leakedinto the bot- tom. The exact form of the IM has now been reexamined and restructured A simple,time-domain split-step algorithm for pulsepropagation in a so as to be computabledirectly. In fact, the convergenceof this restruc- refractingmedium is derivedfrom the time-dependentwave equation turedform, which tracks multiple reflected plane wave spectra, improves (TDWE). The derivationis similarin form and spiritto the factored with slopeangle, thereby making it complementaryto the small slope Helmholtzequation methods used in findinggeneralized parabolic equa- approximatedalgorithm. Calculationof the completeIM removesthe tions.Like theHelmholtz equation analyses, the factored TDWE involves defectsnoted above and thereby establishes the IM firmlyas the analytical nonlocalpseudodifferential operators and a commutatorterm. Making waveobject that describesthe experimentally observed wedge modes. Nu- useof causality,and consideringmedia in whichthe index-of-refraction mericalcomparisons of theIM with wedgesolutions obtained by thepara- changesonly slowly over a typicalacoustic wavelength, the commutator bolicequation and coupledmode codes confirm this conclusion.[Work termcan be neglected and the TDWE reducesto a Schrodinger-likeequa- supportedby ONR OceanAcoustics. ] tionin space-time.From here,a varietyof methods(for example,phase- spacepath integrals)can be employedto generatemarching algorithms for computingthe evolutionof an acousticpulse in a refractingmedium. A fast split-stepalgorithm (of the form FFr/multiply/FFT/multiply/ 9:35 etc.) only resultsif theoperator playing the roleof the Hamiltonianin the pulseSchrodinger equation is separable, which is only a goodapproxima- AAAS. An integral equation approachto range-dependentboundary tionfor narrow-band,forward propagation. However, a somewhatslower value problemsin ocean acoustics.Peter H. Dahl (MIT/WHOI Joint phase-spacealgorithm baseddirectly on the phase-spacepath integral Program in Oceanography/OceanographicEngineering, Woods Hole doesnot sufferany of theselimitations. The time-domainsplit-step algo- OceanographicInstitution, Woods Hole, MA 02543) and George rithm will be comparedwith other pulse-propagationmethods (such as V. Frisk (Department of Ocean Engineering, Woods Hole direct Fourier synthesis)and the resultingaccuracy and numericaleffi- OceanographicInstitution, Woods Hole, MA 02543) ciencies discussed.
Sl 17 J. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America Sl 17 10:20 Thedevelopment and application of microscopic phase space methods andglobal path integral constructions to bothdirect and inverse problems AAAg. Computingocean pulse propagation using parabolic equations and in acousticsare reviewed.The Helmholtzequation is consideredfor ( 1) numerical phase correction. Robert P. Gilbert (Department of one-waydirect wave propagation, (2) two-waydirect wave propagation, MathematicalSciences, University of Delaware,Newark, DE 19716)and and (3) inverseformulations. The one-waytheory concentrates on the David H. Wood (Code 3332, New London Laboratory, Naval constructionof exact,numerical, and uniformperturbation solutions to UnderwaterSystems Center, New London,CT 06320) theWeyl composition equation and their application to thefurther devel- Whenone uses a parabolicequation instead of theHelmholtz equation opmentand testing of thephase space marching algorithm. Specific com- to synthesizea pulsepropagating in the ocean,the so-called"phase er- parisonswill be madewith the IFD codesfor one-waywave equations rors"in solutionsof theparabolic equation lead to errorsin theshape and derivedfrom formal operator rational approximations. The two-waythe- arrivaltime of the pulse.Of course,different kinds of parabolicequations ory considersthe evaluationof the Feynman/Garrodapproximate path are knownto havedifferent types and sizes of phaseerrors. A transmuta- integral,the constructionof an explicitpath integralrepresentation for tion approachis used,that is, the kernelof an integraltransformation is the homogeneousHelmholtz equation, and the combinationof transmu- soughtthat maps solutions of theparabolic equation into solutions of the tationand quantum field theoretic variable-dimensionality ideas to incor- Helmholtzequation. Naturally, different types of parabolicequations re- poratebackscatter. The inverseanalysis seeks to exploitthe Fourierinte- quiredifferent transmutation kernels. Using the appropriate transmuta- gral operatorstructure of both the finite and infinitesimalpropagators tion kernel,the effectsof parabolicphase errors can be numericallycor- andthe semigroup property of theone-way Helmholtz equation to exactly rectedfor. In an oceanwith slow horizontalvariation, it is possibleto solvethe refractiveindex profile reconstruction problem for transversely periodicallyupdate the transmutationkernel as the solutionis tracked inhomogeneousenvironments. [Work supportedby NSF, AFOSR, horizontally.Numerical examples will be presented. ONR. ]
10:35
AAA9. Computingideal starting fields for parabolicequations: Approach 11:20 and examples.E. Richard Robinson and David H. Wood (Code 3122, New London Laboratory, Naval Underwater SystemsCenter, New AAA12. The effects of dispersion,attenuation, range, and other factors London, CT 06320) onthedurationoftransients. DonaldR. DelBalzo, RobertL. Field, and MichaelD. Collins(Naval OceanResearch and Development Activity, A parabolicequation model requires a startingfield, that is, the values NSTL, MS 39529) of the fieldmust be givenas a functionof depthat a fixedrange. The ideal startingfield is a weightedcombination of propagatingnormal modes. In thispaper, the resultsof a theoreticalinvestigation into the effects Sinceit istime consuming to computenormal modes, the usual practice is of variousfactors including dispersion, attenuation, and range on thedu- to usea more easilycomputed result such as a Gaussian,filtered Gaus- ration of transientsin shallowwater oceanenvironments are presented. sian,sine, or uniformocean starting field. However, in our approach,less An acousticsource with the time historys (t;to) = 0 for t < 0 and sin (tot) effortis requiredbecause the individualnormal modes are not computed for t > 0 is placedin the oceanseveral kilometers from a receiver.The duration of the transient is the amount of time between the first arrival but only the desiredcombination of normalmodes that givesthe ideal startingfield. Two examplenormal modestarting fields are computed and the achievementof a time harmonicstate at the receiver.Using. a and are comparedto otherstarting fields. The effectsof the variousstart- range-dependenttime domainmodel developed at NORDA, the Klein- ingfields on thepropagation loss predicted by a parabolicequation model Gordon equationis solvedto obtain the receivedsignal P(t;v) due to the are shown. Gaussiansource S(t;v) = exp[-- (vt)2]. Then,P(t;v) is convolvedto obtainthe reponsep(t;to) dueto s(t;to).Qualitatively, one would expect transientduration to increasewith dispersionand rangeand to decrease with attenuation.A presentationof quantitativeresults to illustratethese 10:50 relationshipsand discussthe possibilityof the useof transientduration to AAAI0. A wide-angleinitial field for parabolicequation models. David solveinverse problems will be given. [Work supportedby ONR and J. Thomson (Defense ResearchEstablishment Pacific, FMO Victoria, NORDA.] BritishColumbia V0S 1B0,Canada) and C. Scan Bohun( Departmentof Physics,University of Victoria,Victoria, British Columbia vgw 2Y2, Canada)
Numericalpredictions of underwatersound propagation are routinely 11:35 carriedout by applying marching algorithms to solve parabolic equations. Theseequations require initial fielddata to beginmarching process. For AAAI3. Transient acousticwave propagationin a fluid layer with manyapplications, a suitablynormalized Gaussian or sinefunction of stepwise range dependent perfectly reflecting boundaries. Gary depth(z) is used[E. R. Robinsonand D. H. Wood,J. Acoust.Soc. Am. H. Brooke, Roderick E. Glover (Defence Research Establishment Suppl.1 81, S10 (1987)]. Thesefunctions, however, do not modelcor- Pacific,FMO Victoria, British ColumbiaV0S 1B0, Canada), and Jacques rectlythe vertical-wavenumber (k:) spectrumassociated with the farfield R. Chamuel(SonGquest/Advanced Ultrasonic Research, P.O. Box 153, of a pointsource. In this paper,a new startingfield is presentedthat WellesleyHills, MA 02181) matchesthe properk, variationover the full spectralaperture needed by wide-angleparabolic equation models. The newfield in thez domainis The coupledmode solution technique JR. B. Evans,]. Acoust.Soc. obtainednumerically via a discreteFourier transform of its bandlimited Am. 74, 188-195(1983) ] isadapted to theproblem of transientacoustic spectrum.Because of thisefficient distribution of energy in thek: domain, wavepropagation in a singlefluid layer with perfectlyreflecting boundar- a largerstep-size (Az) canbe usedthan that requiredby the Gaussian ies.Properties of oneof the boundariesare allowed to varyin a stepwise initial field [H. M. Garon, J. S. Hanna, and P. V. Rost, J. Acoust.Soc. rangedependent manner. The transient response due to animpulsive line Am.Suppl. 161, 812 (1977)]. Numericalexamples comparing the effects sourceis synthesizedfrom solutions to a globalmatrix formulation of the of differentstarting fields on wide-angleparabolic equation predictions coupledmode problem for two-dimensionalwaveguides. Several numeri- are presented. calexamples are presented that model: (i) waveguidescattering behavior ofboth single and multiple scatterers placed near the boundary of thefluid layer and (ii) propagationin a waveguidewith a variable impedance 11:05 boundarycomprised of alternatefree and rigid strips.The numericalre- sultsare comparedwith experimentalobservations taken from measure- AAAll. Phase space and path integral methods in computational mentsperformed at ultrasonicfrequencies using an air-suspendedwave- acoustics.Louis Fishman (Departmentof Mathematics,Colorado guide[J. R. Chamueland O. H. Brooke,J. Acoust.Soc. Am. Suppl.1 77, Schoolof Mines, Golden, CO 80401) Sl3 (1985)].
Sl 18 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115th Meeting:Acoustical Society of America S118 11:50 targetsfor over-the-horizonlong range weapons systems. While acoustic predictionsbased on XBT inputsare useful,they can he improvedby AAA14. In situ acousticmeasurements for tacticalASW applications. supplementingthem with in situacoustic measurements at sea.The infer- CharlesH. Wiseman(Lockheed Missiles & SpaceCompany, Inc., encesas to soundpropagation conditions based only on XBTs arelimited AstronauticsDivision--Advanced Marine Systems,P.O. Box 3504, by their numberand placementconsidered against the backdrop of inho- Sunnyvale,CA 94088-3504) mogeneitiesof the ocean.On the otherhand, the in situ measurementof Te4nperatureversus depth profiles provided by expendablebathy- soundis notan inferenceof soundpropagation--in point of fact,it is the thermograph(XBT), togetherwith informationfrom atlases of seasonal direct agent,sound itself. The ASW forcesshould steal a leaf from the waterconditions and ocean bottom mapping, constitute the inputsfor RadioCommunicator's notebook. While predictions of hf radiopropaga- predictingsound propagation conditions for operationalASW use.At tionare set forth in specialatlases, in situmeasurements of radio propaga- sea,XBT inputsfeed computer models and nomograms that furnish infer- tionconditions in theform of ionosphericsoundings (ionograms) furnish enceas to soundpropagation conditions on whichthe ASW tacticscan be the mostaccurate, real-time indications of usablefrequencies with mini- based--forexample, the routing of convoysor navaltask units to mini- mum multipathdistortions. There are a varietyof methods,overt and mize the likelihoodof detectionby enemysubmarines; the selectionof covert, which can be usedby shipsand aircraft to effectin situ acoustic optimummodes for activeand passivesonars; or the identificationof measurementand these are discussed in the bodyof the paper.
FRIDAY MORNING, 20 MAY 1988 ASPEN ROOM, 8:45 TO 11:35 A.M.
SessionBBB. Musical AcousticsV: General Topicsin Musical Acoustics
Uwe J. Hansen, Chairman Departmentof Physics,Indiana State University,Terre Haute, Indiana 47809
Chairman's Introduetion4:45
ContributedPapers
8:50 soundfield is comparedto the calculatedand measuredforce on the bridgewhen the stringis plucked. BBB1. Dependenceof vibrational modesof a handbellon thicknessand addedmass. H. John Sathoff,Jiang Zhiqing, and Thomas D. Rossing (Department of Physics,Northern Illinois University, DeKalb, IL 60115) 9:20
The vibrationalmodes of a largetuned handbell (G2) are compared BBB3. On the structural-aeoustiedesign ofguitars. EvanB. Davis (3517 with thoseof an untunedbell casting having the same shape but with twice N. E. 6th Street, Renton, WA 98056) thethickness. Doubling the wall thicknessis found to raisethe frequencies of the variousmodes by factorsranging from 1.4 to 2.4 (fiat platetheory A structural-acousticmodel of a guitartop platebacked by a rigid predictsa factorof 2 for all modes).Unlike the tunedG2 handbell,whose cavityhas been built. Acousticcoupling between the interior and exterior pitchcoincides with thefrequency of thefundamental mode of vibration, acousticfields is accomplishedwith a simplemodel of the soundhole. the thickbell has two prominentstrike notes, depending upon where it is Acousticpower spectra are calculatedin 1/12 octaveintervals from 55 to struck.These subjective tones appear to be createdby groupsof three 440Hz. Thesoundboard ismodeled as an orthotropic damped plate with nearlyharmonic partials radiated by the (m,l) and (m,,•) familiesof reinforcingbeams and a circularcutout forced by harmonicpoint loads modes,respectively. The acousticaleffects of attackingup to twelve35-g representingthe strings.The modelrequires the plateoutline to be axi- brasscylinders have also been observed, both in a symmetricalpattern and symmetric.There are no limitationson the numberor orientationof the in clusters similar to the clusters of mei found on ancient Chinese bells reinforcingbeams. Material propertyvariations, brace cross section, [Rossinget al., J. Acoust.Soc. Am. Suppl. I 77, S102 (1985)]. The bracepattern layouts, individual brace tapers, plate thickness, and var- additionalmasses lower the frequenciesof certain modesbut have rela- nishand glueline effectscan be studiedwith this model.A suitabledata- tively little effecton modaldecay rates. baseof materialproperties and guitar geometryparameters, string lengths,bridge placements, etc. has been built to supportthe model.
9:05 9:35
BBB2. Sound radiation and frequencyrealmrise of classicaland folk BBB4. A model analysis of force-deformation characteristics of a piano guitars.Erie T. Watsonand ThomasD. Rossing(Northern Illinois hammer. Hideo Suzuki (Oho Sokki Co., Ltd., 2-4-1 Nishishinjuku University,Department of Physics,DeKalb, IL 60115) Shinjuku-ku,Tokyo 163,Japan ) The soundradiation has been measured as a functionof anglein an Effectsof a geometryand a Young'smodulus profile of a pianoham- anechoieroom when the guitar is drivenwith a sinusoidalforce. Classical mer on its static force-deformationcharacteristics are studied usinga guitarsradiate more stronglythan folk guitarswhen the sameforce is finiteelement mode. The materialproperties of the modelare assumedto appliedto the bridge[J. Poppand T. D. Rossirig,International Sympo- be linearas well as isotropic.Therefore, only a geometricnonlinearity is siumon MusicalAcoustics, Hartford, July 1986]. Radiationconstants considered.A hammer model with a representativegeometry and are largeas 2.7 Pa/N are observedat guitar resonances.Near certain Young'smodulus profile of an actualbass hammer has a force (/)-defor- resonances,dipo!ar or quadrupolarradiation predominates. The radiated mation(d) relationshipapproximated by a function,f= Ads, withap-
S119 J. Acoust.Soc. Am. Suppl. 1, VoL83, SpringI g88 115thMeeting: Acoustical Society of America $119 proximatelyB = 1.7.On theother hand, the degree of nonlinearity(B) of 10:35 a particularhammer obtained from a dynamicmeasurement is approxi- BBBg. On the bore-to-driver feedback in double-reed conical woodwinds. mately2.3 or larger.A hammerwith a largerincrease of Young'smodulus in the inwarddirection has a largerdegree of nonlinearity.The geometry J. Agul16,A. Barjau, J. Martinez, and S. Cardona (Departmentof of a hammerhas a largeeffect on thevalue A in theformula shown above. Mechanical Engineering. Politechnical University of Catalunya, A hammerwith a shorterheight (distance from the top of the hammerto Diagonal,647, 08028 Barcelona,Spain) thetop of thewood piece) or with a largerwidth has a largervalue of A (in The nonlinear behavior of double-reed drivers makes advisable tile use otherwords, a largerstiffness). of time domain to study bore-to-driverfeedback. The characteristic trendsof theimpulse response of a conicalwoodwind are revised. The low intensityof thereflection coming from the first open hole shows the essen- 9:50 tial rolethat Bernoullipressure plays in the reedself-sustained oscillation (alsoproved by the reedsqueaks that playersproduce to checkreed readi- BBBS. An alternate fairy tale for brassinstruments. R. Dean Ayers ness). Time-domain study of theseinstruments leads to the convolution (Departmentof Physics-Astronomy,California State University, Long productrelating pressure and velocity at thebore input section. The useof Beach, CA 90840) the impulseresponse h (t), the plane-waveand the spherical-wavereflec- tionfunctions, Rp (t) andRs (t) respectively,askernel functions in this Arthur Benade'swell-known paperback Horns, Strings, andHarmony productis revised.The sharpfluctuations of Bernoullipressure may im- includesa didacticfairy tale abouta buglerforced to playon a uniform pair the useof Rp(t) whileR• (t) is unsuitabledue to the usualtaper pipeand stayin tunewith ordinarybugles. With a bit of sawingand some discontinuityin the staple-mainbody junction of theseinstruments. Com- impressivelipping, he managesto succeed.This storyfits nicely with the putedand experimentalinitial transientsare givenfor the caseof a "ten- prevailingview that the bell on a brassinstrument causes a considerable ora" (Catalan tenor shawre). variationin theeffective length of a (singly)closed pipe, thus making the performer'sjob mucheasier. Another possible ending for thisfairy taleis that the buglersolves his tuning problems with the hand-stoppingtech- niqueof naturalhorn players, rather than with hisembouchure. One day heaccidentally closes the pipe completely and discovers that he now has a perfectlygood harmonic series. A slightlengthening of the pipeputs him 10:50 in tune with his colleagues.This endingis consistentwith a viewjust BBB9. Woodwind instrument simulation in real-time. Michael Park and emergingfrom time-domainstudies that, in termsof reflectionback to the DouglasH. Keefe(Systematic Musicology Program, School of Music, dri•er, the bell of a brassinstrument mimics the closedend of a uniform DN-10, Universityof Washington,Seattle, WA 98195) pipe.Demonstrations will be included. A real-time implementationof the Mclntyre-Schumacher-Wood- housetime-domain description of musicaloscillators (1983) has been developedusing digital signal processing (DSP) hardware.The systemis 10:05 comprisedof a microcomputerwith installed DSP board including a digi- tal-to-analogconverter. The taskis to performa convolutionand solve BBB6. Influenceof bore shapeon the resonantfrequencies and spectral twosimultaneous equations with nonlinearities for the unknown pressure envelopeof a Frenchhorn. J. Duane Dudley and William J. Strong andvolume flow at successiveinstants of time.A tablelookup with linear (Departmentof Physicsand Astronomy,Brigham Young University, Provo, UT 84602) interpolationsimplifies the originalformulation. The reflectionfunctions usedin the convolutionare calculatedbased upon analytical models of The harmonicityof the resonantfrequencies, and the shapeof the woodwinds,but the nonlinearvalving mechanism is somewhat idealized. spectralenvelope of the inputimpedance curve, are two importantindica- One can "play" the instrumentvia an MIDI keyboardcontroller. Each torsthat mightbe used in predictingthe usefulnessof a givenair column MIDI note-onmessage loads a differentreflection function into the DSP asa brasswind instrument. These can be determinedby approximating boardmemory, while the pitchwheel controls "blowing pressure." Re- the air columnas a largenumber of concatenated1ossy cylindrical incre- cordedexamples will be presented. ments,and computing and plotting the input impedance of theconfigura- tion asa functionof frequency.In thispaper the resultsof suchcalcula- tionsare described for a numberof boreshapes related to theFrench horn, andthe effectsof changingvarious dimensional parameters are illustrat- ed. The contributionsof the bell,the leadpipe,and the mouthpieceof a typicalopen B-flat horn are firstexamined. Then thehorn isidealized and 11:05 representedby 14 dimensionalparameters, so that each can be varied independentlyto studyits effectson harmonicityand spectral envelope, BBBI0. Patternsof expressivetiming in performancesof a Beethoven andan optimumconfiguration isdetermined. Finally, the effects of insert- Minuetby nineteenfamous pianists and one computer. Bruno H. Repp ing the hand into the bell are examined. (HaskinsLaboratories, 270 CrownStreet, New Haven,CT 06511-6695) Onset-to-onsetdurations of quarternotes and sixteenthnotes were 'measuredinnineteen complete performances ofthe third movement of Beethoven'sPiano Sonata op. 31, No. 3. Thesemeasurements were com- 10:20 paredto thosefrom a computerperformance incorporating the "Beetho- BBB7. Wind instrumenttransfer responses.James W. Beauchamp venpulse" expressive microstructure devised by ManfredClynes [U.S. (School of Music, University of Illinois at Urbana-Champaign, Urbana, PatentNo. 4,70•,682 (1987) ]. This pulseis a particularpattern of rela- IL 61801 ) tive notedurations (and amplitudes)within successivetime units,and it isdesigned to beinvariant except for occasionallarger deviations at struc- Mouthpieceand output pressuresignals from severalinstruments tarallysalient points. Although patterns resembling the Beethoven pulse (e.g., trombone,clarinet, saxophone)have been reoorded,harmonic couldbe found locally in thehuman performances, on the whole there was spectrahave been computed, and transfer responses have been estimated littleevidence for a fixed,pulselike pattern of timingdeviations. Rather, by taking ratios of correspondingharmonic amplitudes. Each transfer themeasurements suggest that great artists change the patterns of relative responseappears to be a seriesof pointson a smoothhigh passfilter notedurations continuously to meet the local expressive requirements of function,even though it actuallycorresponds to pointson (tendingto- the composition,and that, despiteconsiderable individual differences, ward the minimaof) a multiresonantresponse curve. According to sim- thereis someconsensus about what these requirements are. If theinvar- plifiedassumptions the measuredresponses would be independent of am- iantBeethoven pulse enhances the expressiveness of the computer perfor- plitude and fundamental frequency. However, there is, in fact, a mance,as it seemsto do [B. H. Repp,J. Acoust. Soc. Am. Suppl. 181, S92 considerabledegree of variation.Results will be presentedand reasonsfor (1987)], thisis apparently not because it captures aspects of humanper- the variations will be discussed. formance.[Work supportedby NIH.]
S120 d. Acoust.Soc. Am. Suppl. 1, Vol.83, Spring1988 115thMeeting: Acoustical Society of America S120 11:20 bridge.The resultingsound is referredto by Indianmusicians as a buzz. Withoutjuari the pitch heardis veryclose to the stringfundamental and BBBll. The role of tambura spectrain drone tuningsof North Indian the partialsfall linearlyin decibels.With juari the pitchis heardat oneor' ragas.Edward C. Carterette,Nazir Jairazbhoy,and Kathryn Vaughn eventwo octavesabove the basepitch and the partials have great power as (Departmentsof Psychologyand Music, Universityof California,Los far asthe 20th partial. With juari the 4th, 7th, 11th, or 12th,and the 17th Angeles,CA 90024) partialshave more amplitude than the fundamental.Total harmonicdis- As part of a programon the role of drone interactionswith North tortionis on the orderof 1 to 19dB withoutjuari but rangesfrom 12 to 44 Indian ragasplayed on the sitar, the waveformsof the tamburafor the dB with juari. The upwardspread of energyinto higherpartials imparts mostcommon drone tuning of thefour strings[pa, sal, sa2,sa' (low sa) ] richnessto tamburatones, and underliesthe useof differentdrone tunings wereanalyzed. Each of the stringswas playedwith and withoutjuari for differentragas. The pitchand timbre changes with tamburaare dis- threadsthat havethe effectof separatingthe stringslightly from the cussedin termsof the psychoacousticsof complex tones.
FRIDAY MORNING, 20 MAY 1988 GRAND BALLROOM A, 9:00 TO 11:00 A.M.
SessionCCC. PsychologicalAcoustics VII: Multiear Phenomenaand Models
Honor O'Malley, Chairman TeachersCollege, Columbia University, 525 W. 120thStreet, New York,New York 10027
ContributedPapers
9:00 9:30
CCC1. Creating a sound field for manilmlating interaural level CCC3. Binaural modulationdetection. D. Wesley Grantham and Sid differencesseparately or independentlyof the interauraltime differences P. Bacon (Bill WilkersonHearing & SpeechCenter and Divisionof and vice versa.George F. Kuhn (VibrasoundResearch Corporation, Hearingand Speech Sciences, Vanderbilt University, 1114 19th Avenue 2855West Oxford Avenue, Englewood, CO 80110) South, Nashville, TN 37212) Usingtwo or moresources whose source strengths and relative phases A sinusoidallyamplitude-modulated (SAM) noise (modulation canbe controlled independently, a sound field can be generated in which depth:1.0) wasused to maskanother SAM noise,the signal, in a two- the interauralpressure level differences (ILDs) andthe interauraltime intervalforced-choice experiment. Signal and maskermodulation fre- differences(ITDs) canbe controlled separately. Thus, for a fixedILD, a quencywere both either 4 or 32 Hz, andwere added in quadrature.The rangeof ITDs, or viceversa, can be presentedto the listener.Thus it samewideband noise carrier was used for both maskerand signal(spec- appearspossible to performsome localization experiments that had pre- trum level:15 dB SPL). The wholestimulus was filtered after modulation viouslybeen confined to lateralizationmethods using earphones. Some (either4-kHz low pass or 4-kHz high pass). Modulation depth (m) ofthe calculated results for interaural time differences and interaural level dif- signalwas varied adaptively to trackthreshold. In thecase of dioticpre- ferences,using two sources,will be presented. sentation(Mo-So), thresholdm was about0.30 for both modulation frequenciesand both filtering conditions. In a binauralcondition Mo-Sm (maskerpresented to bothears and signal to onlyone ear) some subjects. showeda releasefrom masking of up to 10dB (m = 0.10), whileothers. showednone. The release from masking occurred for bothfiltering condi- tions,but onlyfor the lower(4-Hz) modulationrate. The resultswill be, 9:15 discussedin relationto currentmodels of binauralprocessing and the CCC2. An adaptationmodel for sourcedisplacement in multiple-source sluggishnature of thebinaural system. [Work supported by NIH.] environments.Gregory H. Wakefield (Department of Electrical Engineeringand Computer Science, University of Michigan,Ann Arbor, MI 48109-1109)
In general,the acoustic field at thetwo ears is notsufficient to support 9:45 sourcelocalization for morethan one source. Experiments on localization indicate that the co-occurrence of two sources can lead to substantial CCCA. Detection of interaural phaseshifts and level differencesin sourcelocalization error that reflectseither the ambiguitypresent in the I•ortions of broadbandnoise. William A. Yost (Parrely Hearing acousticfield or someform of biasin the binauralprocessing of that Institute,Loyola University, 6525 N. SheridanRoad, Chicago, IL 60626) acousticfield. It is shownthat displacementof the perceivedlateral posi- tion of a narrow-bandnoise away from the lateralposition of a narrow- Versionsof Cramer-Hugginspitch stimuli were generated such that a band interferer [G. H. Wakefieldand N. F. Viemeister,J. Acoust.Soc. narrowband (bandwidths ranged from 1-100 Hz) of a wideband,diotic Am. Suppl.176, S91 (1984) ] doesnot lie on the surface of possiblesource noise(low-passed filtered at 8000Hz) waspresented dichotically with an positionssupported by the acousticfield. Therefore, it doesnot appear interauralphase shift or an interaurallevel difference. The threshold val- that anomalouslocalization can be explained entirely with respectto am- uefor detectinga phaseshift or a leveldifference in thedichotic, narrow.- biguitiesin the acousticfield. Instead, a modificationof Stern'sposition bandportion of thebroadband noise was determined as a functionof the variableis proposedto accountfor sourcedisplacement by introducinga centerfrequency of the dichotic, narrow band of noise and as a functionof frequency-dependentform of spatialadaptation. Much, but not all, of our the bandwidthof the dichotic,narrow band of noise.A single-interval displacementdata are consistent with an additive form of adaptation with procedurewithout feedback was used to estimatethree point psychomet.- a timeconstant of 25 msin whichthe levelof adaptationat a givenposition ric functionsrelating d' to interauralphase or d' to interaurallevel differ- doesnot depend on eftcitation at otherpositions. The manner in whichthe ence. A random variation in the interaural level difference of the entire modelfails to predictdisplacement for long-durationsources suggests the noisewaveform was also used to forcethe listenersto usepitch as the cut: presenceof a weakeradaptation across position. for detectionrather than some aspect of lateralposition. The datawill be
S121 J. Acoust.Soc. Am. Suppl.1, Vol. 83, Spring1988 115th Meeting:Acoustical Society of America S1;!1 discussedin termsof modelsof binauralprocessing and binaural pitch. 10:30 [ Work supportedby NINCDS. ] CCC7. Discriminationof directionin the free field. Ervin R. Halter, Thomas N. Buell, David A. Basiji, and Elizabeth E. Shriberg (Departmentof Psychology,University of California,Berkeley, CA 94720) 10:00 The psychophysicsof direct comparison(two-alternative, forced CC-'•5. Laternli-•tion on the basisof interaural envelopedelays: The choice)was applied to thestudy of localization.For complexsignals, the effectof componentstarting phase. R. H. Dye, A. J. Niemiec,and M. interauralparameters presented by stimuliin the freefield are more"nat- A. Stellmack(Parrely Heating Institute, Loyola University of Chicago, ural" than thoseheard through earphones though more difficult to con- 6525 N. SheridanRoad, Chicago, IL 60626) trol.In thisstudy, digital preprocessing was used to reversethe spectral Thresholdinteraural envelope delays were measured with a 2 AFC characteristicsof each loudspeaker to makeits responseuniform and in- paradigmas a functionof modulationfrequency L/e,, = 25, 50, 100,200, discriminable. The listener sat in an anechoic chamber with his or her 300,400, and 500 Hz) for three-and five-component complexes whose headoptically fixed at thecenter of a semicirculararray of speakers.The centerfrequencies (f•) wereequal to 4000 Hz. Comparisonswere made stimuli were trains of n filtered clicks (n = 1, 9, or 18) whosecenter betweenthresholds obtained when the starting phases were randomized frequencieswere either 2050, 4050, or 6050 Hz. Interclick intervals andwhen they were all setto zero.The levelof eachcomponent was (ICIs) wereeither 4 or 12ms and levels were set so that a singleclick was approximately50 dB SPL, and the signal duration was 200 ms with 10-ms l0 dB SL. Minimumaudible angles (MAAs) werefound to increaseas linearrise/decay times. While reductionsin depthof modulationelevate theazimuth increased from 0 to 60 deg;performance was best at 2050Hz. thresholdenvelope delays [O. B. Henning,J. Acoust.Soc. Am. 55, 84-90 Studiesof lateralizationwith earphoneshave led to speculationsabout (1974) ], no sucheffects were found for phaserandomization even when "binauraladaptation," a processwhereby improvements in threshold modulationrates were low andall componentsfell within a criticalband. withincreased n depend on the click rate ( 1/ICI). Similaranalyses show Thisseems surprising in lightof the fact that both reduced depth of modu- thisalso to be truefor free-fieldlistening. [Research supported by the lationand phase randomization reduce the peak factors of thewaveforms, NIH.] whichis commonlybelieved to interferewith entrainmentby the peri- pheralauditory system. Efforts to reconcilethe absenceof phaseeffects with the deleteriouseffects of reduceddepth of modulationwill be pre- sented.[Work supportedby NIH/NINCDS andAFOSR.] 10:45 CCCS. Sound inealizationwithout interaural time or intensity differences.Frederic Wightman, Doris Kistler,Marianne Arruda,and Linda Schwister(Department of Psychologyand WaismanCenter, 10:15 Universityof Wisconsin,Madison, WI 53706)
CCC6. Accruingbinaural cues across frequency. Thomas N. Buelland In a seriesof localizationexperiments, blindfolded listeners were Ervin R. Hafter (PsychologyDepartment, University of California, askedto givethe apparentazimuth and elevationcoordinates of 250-ms Berkeley,CA 94720) broadbandnoise bursts. The noisebursts were spectrally shaped so that thespectrum level in eachcritical band was set randomly (from trial to Localizinga complexsound requires in part an analysisof the inter- trial) withina 20-dBrange. In a referencecondition, the soundswere auraldifference of time (IDT) of the componentfrequencies. Above transducedby miniatureloudspeakers placed at 36 locationsin an ane- about500 Hz, the IDT is constantacross the frequencyspectrum. This tholechamber. Performance in this free-field condition agreed with pre- commonalitymay be usedto groupcomponents that arisefrom a single viousresults in theliterature except for anincreased frequency of front- sourceas well as segregatingsuch components from thosewhich arise back reversals,which is felt to be the result of the lack of a visualframe of fromcompeting sources. Interactions across components can be shown by reference.In a controlcondition, digital techniques were used to synthe- measuringchanges in IDT thresholdswhen a targetis lateralizedin the sizeheadphone-presented stimuli that were nearly identical, as measured presenceof othersimultaneous sounds. When the nontarget sound has a in listeners'ear canals, to thefree-field stimuli. Localization performance dissimilarIDT, thresholdsmay actuallyincrease. This effecthas been withthese stimuli was virtually the same as in freefield, thus verifying the aptly termed"interference" [D. McFaddenand E.G. Pasanen,J. Acoust. adequacyof thesimulation. The digitalfilters used to synthesizethe stim- Soc.Am. $9, 634-639 (1976) ]. In the presentexperiment, listeners de- uli were then modified such that in one condition the interaural time tectedan IDT in a 600-Hztarget tone both with andwithout the presence differenceswere removed from all 36 stimuli,and in anothercondition the of a nontargettone. Of specialinterest was the harmonic relation between interauralamplitude differences were removed. In both of thesecondi- targetand nontarget. Among other factors, the results show that a simple tions,localization performance was degraded. However• preliminary re- relationsuch as 400 Hz can producemore interferencethan one that is sultssuggest that the degradationcaused by removalof interauraltime nonharmonic,even when the interfering tone is closerto thetarget, e.g., differencesis considerably greater than that caused by removalof inter- 531 Hz. [Researchsupported by NIH.] auralintensity differences. [Work supported by NIH, NSF, andNASA. ]
8122 J. Acoust.Sec. Am. Suppl. 1, Vol. 83, Spring 1988 115thMeeting: Acoustical Society of America S122