Measurement of Velocity Profiles of the Jets Issuing from Some Flue Geometries Typical of Air-Jet Instruments Shigeru Yoshikawa and Keita Arimoto Dept. of Acoustical Design, Kyushu Institute of Design, Fukuoka, 815-8540 Japan

Air-jet musical instruments may be categorized by the geometry of flue channel and flue exit: The flue of metal organ pipes is simply modelled by a vertical plate and a horizontal languid (such a flue is called "organ" here). The flue made by flute or shakuhachi players may be modelled by two thin plates corresponding to player's lips (called "short" because of short channel length). Contrary to this "short" flue, a "long" flue consisting of two long plates has been used for experimental organ pipe models. The recorder flue can be modelled as a long flue with a chamfer on the edge of the lower plate (called "chamfer"). The measurement of jet velocity profile was carried out on these flue models without using pipe resonators. The profile was measured at the distances of 2 to 25 mm from the exit when the initial jet velocity was varied from 10 to 50 m/s. The flue height was 2.2mm throughout the measurement. The profile difference between the “short” and “long” flues is distinctive as inferred from the top-hat and Poiseuille profiles at the exit. The profile from the "chamfer" flue tends to change from the "long"-type profile for lower jet velocities to the "short"-type one for higher velocities. The "organ" flue indicates an asymmetric profile for shorter distances from the exit. The "organ" jet directs downwards when the languid is 5 mm thick, while directs upwards when it is 1 mm thick. The experimental results are compared with the theoretical Bickley profile on a laminar two-dimensional jet and an empirical Nolle profile more squarish than the Bickley profile.

INTRODUCTION anemometer when the probe is precisely dislocated with a 3-D adjustable traverser. The measurement parameters were set as follows : Tones of air-jet instruments are considerably Initial center line velocity (U00) 10, 20, 40, 50 m/s affected by the jet velocity profile which depends on Distance from the flue exit (x) 2, 4, 8, 15, 25 mm the geometry of the flue channel and flue exit. The transverse velocity profile of the jet was measured However, there are very few reports that give for the fixed U00 and x values. quantitative measurement of the velocity profile or some correlation between the flue geometry and the 2h foot 98mm foot 2h velocity profile. Our aim is to make the comparison of 100mm the velocity profiles resulting from some flue 5mm geometries typical of air-jet instruments such as the (a) long (b) short pipe organ, flute, and recorder. 2mm 2h 2h foot 1mm 5mm, MEASUREMENT 1mm (c) chamfer (enlarged near the flue exit) (d) organ5 , (e) organ1 The flue geometries used in our experiment are Fig.1 Flue geometries used in our experiment. illustrated in Fig. 1 . The “long” has a long and uniform flue channel and flat exit. Such a flue is often used in RESULTS experimental organ pipe models. The “chamfer” has the same flue geometry except for the chamfer on the Figure 2 shows the profiles of the “long” and lower side of flue exit. The flue geometry of “short” “short” obtained at the same experimental condition may give a simplified model of shakuhachi or flute (U00= 10 m/s, x = 4 mm). These profiles are well player’s lips. The “organ5” and “organ1” are different approximated by from each other in the thickness of languid (5mm and U (x, z) = U (x)sech 2 (z / b)n (n = 1,2,3 ⋅ ⋅ ⋅), (1) 1mm). The flue height 2h is 2.2mm in all of the flue 0 geometries throughout the measurement. where U0 defines the centerline jet velocity and b The velocity profile is measured by using a hot-wire the jet half-thickness. The profile from the “long” flue Poiseuille profile

top-hat profile

Fig.2 Velocity profiles of “long” and Fig.3 Channel flow at the flue Fig.4 Velocity profiles for U00 = “short” for U00 =10 m/s and x=4 mm . exit of the “long” and “short” flues. 20 m/s and x= 8 mm.

Fig.5 Velocity profiles for U00 Fig.6 Velocity profiles of “organ5” Fig.7 Velocity profiles of “organ1”

=40 m/s and x=8 mm. for U00 =20 m/s and x=2, 8, 15 mm. for U00 =20 m/s and x=2, 8, 15 mm. has n=1 and shows the so-called Bickley profile; the changes from the “long”-type profile (cf. Fig.4) to the “short”-type profile (cf. Fig.5) when the initial jet “short” flue has n=3 and shows the so-called Nolle velocity U00 increases from 20m/s to 40m/s. profile. The difference between the profiles is inferred (2) The jets of “organ5” and “organ1” do not run from the difference in channel flow at the flue exit, as straight along the x axis as illustrated in Figs.6 and 7. illustrated in Fig.3 . It may be supposed that the More interestingly, the “organ5” jet gradually deviates channel flow has reached the following Poiseuille downwards (about –10 degrees) , while the “organ1” profile at the flue exit of the “long” flue. jet deviates upwards ( about 10 degrees) . (3) Initial velocity profiles (at x= 2 mm) are also U (x, z) = U (x)(h2 − z 2 )/ h2 z ≤ h (2) 0 different between the “organ5” and “organ1” as On the other hand, the top-hat profile may be assumed indicated in Figs. 6 and 7. The “organ5” profile is at the flue exit of the “short” flue. The above asymmetrical and distorted, although it becomes mentioned difference between the velocity profiles symmetrical as the jet travels downstream. The from the “long” and “short” flues is held up to x= 8 “organ1” profile is symmetrical and it is close to the mm and U00 = 20 m/s (the Reynolds’ number Re≈ top-hat profile rather than the Nolle profile. 3000). Figures 4 and 5 indicate the profile difference at x= CONCLUSIONS

8 mm between five flue geometries for U00 = 20 m/s and 40 m/s, respectively. We may easily recognize the Velocity profiles of the jets issuing from five kinds individuality (or separation) of the profiles (although of flue geometries were measured and compared. The the profiles of “chamfer” and “organ1” are partly individuality of five profiles was recognized in the range of 10 ≤ U00 ≤ 20 m/s and 2 ≤ x ≤ 8 mm. overlapping for |z| > 1mm) when U00 = 20 m/s. However, the profiles seem to be divided into two, the When U00 was increased to 40 m/s, four profiles except for the “long” one tended to make up one group, “long” profile and the other profiles when U00 = 40 m/s. Also, all the profiles at x ≥ 15 mm approach to which may be roughly represented by the “short” one. the Bickley profile regardless of flue geometry. Some Finally, all the profiles at x ≥ 15 mm approached to characteristics typical in other flue geometries are the theoretical Bickley profile. Also, significant effects summarized as follows; of the languid thickness were recognized from the (1) The behavior of the jet from the “chamfer” “organ1” and “organ5” profiles. Can wall vibrations alter the sound of a flue organ pipe? M. Kob Institute of Technical Acoustics, Technical University Aachen, D-52056 Aachen, Germany

The prediction of changes in the perceived sound of a blown pipe due to wall vibrations is made difficult by the multitude of interac- tions. Excitation, shape, and sound radiation of structural modes depend on a number of parameters like material, voicing technique, geometry and fixing of the pipe. This article presents experimental work on comparison of vibrations and sound radiation from a tin-rich pipe in two cases: with damped and undamped wall vibrations. It was found out that changes in sound pressure level at certain frequencies in the spectrogram coincide with eigenfrequencies of both air modes and structural modes and thus support the assumption of mode coupling being responsible for sound changes.

INTRODUCTION shown to the right. In this spectrogram, the clouds are still present but the sound pressure level at certain frequencies Although most organ builders agree to organ pipe vi- has been reduced by approx. 10 dB at 1250 Hz, 1550 Hz brations being audible, this is in contradiction to many and 1800 Hz in the first 100 ms of the sound. Smaller experiments that were carried out on modern organ pipes differences between the damped and the undamped pipe (for an overview, see [1]). A reason why this question sound can be observed in the stationary part of the sound is not easy to answer is the multitude of parameters (e.g. at those frequencies. foot pressure, voicing) and boundary conditions (e.g. pipe support, temperature) that are difficult to control during an experiment. In addition, modern flue organ pipes are rather thick-walled compared to pipes of the 17th or 18th century. Stationary sound This work presents some measurement results indi- cating that eigenmodes of the air column, further called As a second approach the pipe was mechanically ex- air modes, and eigenmodes of the pipe body, structure cited with a shaker at the labium (c.f. Fig. 2). The sound modes, are likely to interact at some frequencies. pressure at the upper (passive) end of the pipe has been recorded and the ratio to the applied force has been calcu- lated. For this frequency response functions (FRF) four METHOD cases have been investigated: damped/undamped walls and damped/undamped air column. Several experiments have been carried out (for details, see [1, 2]) for measurement of the air modes and structure In a 3rd experiment, the eigenmodes of the pipe body modes of the same pipe under two different conditions. have been determined from laser velocimetry on the body For detection of the structure modes the pipe was inves- of the mechanically excited pipe and subsequent modal tigated either with walls covered by a removable, heavy analysis. The results are compared to finite element cal- damping layer or without layer. The air modes were iden- culations (FE). The measurement results are listed in Ta- tified by insertion of a paper covered stick into the air ble 1. column inside the pipe. In both cases damping of the res- onances by 10 dB could be achieved. Table 1. Comparison of resonance frequencies (in Hz) from cal- culations (calc.) and measurements (meas.).

Transient sound Structure modes Air modes FE calc. Laser meas. FRF meas. FRF meas. At first the pipe was blown and the sound pressure was recorded. A 829 841 850 896 Figure 1 shows the spectrograms of the pipe in two B 1226 1241 1263 1209 cases. For sake of better visualization, the harmonics C 1514 1500 1516 1524 have been removed. To the left the undamped case is D 1863 1853 1865 1840 shown. Clearly several clouds are visible during the build-up of the sound. The sound of the damped pipe is 50 100 150 200 250 300 50 100 150 200 250 300 t [ms] t [ms] 2500 2500 f [Hz] f [Hz] 2000 2000

1500 1500

1000 1000

500 500

0 0

Hanning window width: 92.8798 ms −40 −20 0 Hanning window width: 92.8798 ms −40 −20 0 Overlap: 2.9025 ms SPL [dB] Overlap: 2.9025 ms SPL [dB]

FIGURE 1. Spectrograms of the undamped (left) and damped (right) pipe without harmonics.

Mic. is bounded by an elliptical tube, vibrating in a twisting Pressure mode. Miklos and Angster [6] observed subharmonics in Force A/D the spectrum and deduced periodic wall stiffening from Amp. the non-linear effect caused by pressure fluctuations in- PC Velocity side the pipe. However, more experiments should vali- with Laser Pipe date these coupling theories. Monkey Forest Vibrometer

D/A Shaker Amp. ACKNOWLEDGMENTS

The author wishes to thank Michael Vorländer and FIGURE 2. Set-up for the FRF measurements. Mendel Kleiner for the opportunity to participate in the GOArt project. The discussions with Vincent Rioux, Pe- ter Svensson and Wolfgang Kropp were invaluable. DISCUSSION

The wall vibrations appear to affect only a small fre- REFERENCES quency range (Modes A-D between 800 Hz to 1900 Hz, as presented in Table 1). Since the resonance frequencies 1. M. Kob, ACUSTICA acta acustica 86, 642-648 (2000). of the structural modes are similar to the eigenfrequen- 2. M. Kob, ACUSTICA acta acustica 86, 755-757 (2000). cies of the air modes, the coupling theory is supported. 3. F. Gautier and N. Tahani, Journal of Sound and Vibration Perceptually, the differences are very small1. 213, 107-125 (1998). In the last two years, some more experiments have 4. C. N. Nederveen and J.-P. Dalmont, “Experimental investi- been carried out that seem to support the organ builders. gations of wall influences on woodwind instrument sound”,

Effort has been made to explain the nature of the coupling in ACUSTICA acta acustica 85 Suppl. 1, 1999, S76. between air modes and structure modes and the hypothe- 5. C. N. Nederveen and J.-P. Dalmont, “Influence of Wall Vi- sis of pipe vibrations being audible. In [3] a mathematical brations on Organ Pipe Sound Level and Pitch”, in Abstract approach to the theory of coupling in a simplified musi- book of The Physics Congress 2000 Physics of Musical cal instrument is presented. Nederveen [4, 5] explains Instruments (POMI), Brighton, 2000, 54-55. the coupling as a change of the compliance of the air that 6. A. Miklos and J. Angster, “Linear and Nonlinear Wall Vi- brations of Organ Pipes”, in Abstract book of The Physics

Congress 2000 Physics of Musical Instruments (POMI), 1 Sound examples and color pictures can be found on the Internet at Brighton, 2000, 54. http://www.akustik.rwth-aachen.de/˜malte/pipe.

¢¡¤£¦¥¨§ ©  © ¡§¡ © ¥ ©¥  ©  ©! " $# ¥ %¡¥'& & ¥(© #)

*¡+ ,%#© $-/.¥ $& 0 ¥ £¥'.1-/ $& ¥(©#)

2%3547698*:<; =?>@; =A6CBEDFHG$3IFHJLK"6CMEDE; NO

OLPHQ?RLSUTWVXQLVCYZXSU[ \5]$^`_bacR(dfeg[ih Tcjkeg[bVmlmn"SodLp"arqs_bl`tvuoaxwLeg[k_kaxVXQLV"PZ[yeza{jgSU[mqs|~}Ieg[b7Qsuou9l{}Ieg[€dfeg[bl‚9Vƒ[g„¢ s|¢†ˆ‡ ‰ Šz‹ŒP[€eza{jgSU[mqs|

eg[b7Qsuo^

Ž

mf‘?’k“‚”–•W—`˜g”–•W™ˆ“™ˆŒš› ”–’k“i•rŒ˜g”–› œ˜gž"Ÿžr˜k f•r —¢¡y™?$£f•¤”–•r™ˆ“i¥5•¤”–¦§ ™?›f”–¦Œ˜ˆ$£'‘`™L¡k˜ˆž"”¨œ˜ˆ¡y”˜gœ–’7ŸU’€œ¨©ª™ˆœ–§ ’k£'¥5•¤”–¦¦™g”¨«{¥5•¤œ–’

˜ˆ ’k§ ™?§ ’y”¨œ¨ I¬x­¯®/°²±v˜g$£Ÿ˜gœ¨”–•W¡kžW’³•W§´˜ˆ—ˆ’<‘?’kžW™L¡k•W§ ’y”¨œ¨ 7¬¶µƒ·²±€¸L¹E¦’<‘`™L¡b˜gž”¨œ˜ˆ¡y”~¡b˜ˆ´ºU’5©ª™gœ–§ ’b£•WI“¨› ¡€¦I˜²¥"˜k ”–™¢¡k™ˆ›Ÿ žr’5”–™

”–¦’Eš› ”–’5˜gœ–™ˆ›£¼»¯½f­<¾?¸Z¿ ™ˆœH”–¦’5˜ˆ¡y™?›“‚”–•W¡5À$’kžr£ ˜7Á{œ–’k“¨™?˜ˆ¡y’k¢’yÃĒk¡y”v•W“v™ˆº“¨’yœ–‘`’k£oÅZ¹E¦ ’5’k‘`’k¦˜gœ–§ ™ˆ•W¡k“v˜gœ–’5“¨›Ÿ Ÿ œ–’k“¨“¨’b£› Ÿ

”–™!˜!©x˜g¡y”–™ˆœ"”m¥E™!•WI”–¦ ’<Ÿfœ–’k“¨“¨› œ–’³˜ˆ§ ŸžW•¤”–›$£f’<› $£f’yœE¡k’€œ¨”˜ˆ•W¼¡y™?$£f•¤”–•r™ˆ“k¸Hm”~•W“E“¨¦™b¥5´”–¦$˜g”~”–¦’¯‘`’kžW™L¡k•¤”ƒ ´™g©C”–¦’Æ‚’€”"•W“"“‚”¨œ–™?—ˆž¤

•W¡yœ–’b˜g“¨’b£Ç£f› œ–•W—¢”–¦’<‘`™ˆœ¨”–’€È “¨¦’b£ £ •W— ˜g”"”–¦ ’³žr˜ˆº •W›§7¸¹E¦ ’²§ ™?›f”–¦´œ–’y“¨™?$˜g¡k’³¡yœ–’b˜g”–’y“5˜ Ÿ¦˜ˆ“¨’<“¨¦•¤©É”5™g©@”–¦’<‘`’kžW™L¡k•¤”ƒ ´ŸU’b˜g½U¸

¹E¦’¢™?º “¨’yœ–‘ˆ˜g”–•W™? “ʘgœ–’!¡yœ–™?“¨“³¡€¦ ’k¡€½?’b£ÇºL ˵ƒ·¤¥5¦ •r¡¦/£ •W“¨Ÿžr˜k f“<”–¦ ’¢”–¦ ’i£f’y”˜ˆ•WžW“³™g©v”–¦ ’¢‘`™ˆœ¨”–’yÈ7˜ˆºU™z‘`’¢”–¦’¢žr˜ˆº •r› §7¸5®Ì•¤”–¦A˜ˆ ’yÈs”–’yœ–$˜gžWžW ¼£sœ–•r‘?’k7˜g¡k™?› “‚”–•r¡k˜ˆžCÀ’kžr£I”–¦’³‘`’kžW™L¡k•¤”ƒ IÀ$’kžr£I’b˜zœ<ºU’€”ƒ¥E’y’k7žr˜ˆº •r› §Í˜ˆ£Iš$› ’²•W“5§ ’b˜ˆ“¨›fœ–’b£I©É™ˆœ<¡k™?§ Ÿ˜gœ–•W“¨™?o¸

light sheet LIPS MOUTH PISTON optics light sheet microphone 2 embouchure air inlet

head flue

1 cm

Î~ÏÉÐfÑÄÒyÓ'ÔfÕÖ Î~ÏÉÐfÑÄÒyÓÇøÄÕ

’y”–› Ÿ¨™ˆ©5”–¦ ’ ’yÈfŸU’yœ–•W§ ’kL”i˜g“i‘s•W’y¥E’k£¨ºs /”–¦’ µœ–’y“¨“¨› œ–’!“¨•r—ˆ$˜gžH™g©”–¦’!¦$˜gœ–§ ™ˆ•W¡k“³˜ˆ“²˜ ©ª› ¡y«

µHm·×¡b˜g§ ’yœ˜i˜gžW™?—”–¦ ’³š$›f”–’?¸ ”–•r™ˆ7™ˆ©Z”–¦ ’²žW’k—g”–¦7™ˆ©C”–¦’²§ ™ˆ› ”–¦¼œ–’y“¨™?$˜z”–™ˆœb¸

ØzÙ'Ú´ÛAÜAÝ¢Þ'߼ڴ؈ÜËÙ

÷

êgæ ôcӈãÄÐsåyáŒæ7ӈçfä€ÑÄÒyӈé¨ýÊÏcå€á'åyáÄÓ ÑÄÏcôªé$ö‚ÏcãæÇϪêzÒyîfòoáÄîfãÄÓ

#

æ7ÏÉêgҀî òÄáÄîfãoÓ ø$bÕ5à²áÄÓ´ä€îfÑoãoé/òoҀÓ?ä€ä€ÑÄÒyÓ îfíHå€áoÓâoÑUå€Ó Ϫä

à²áÄÓÊâoÑoÓ!ÏÉãoä–åyҀÑoæÇӈãåyä¯çfҀÓ!Ïcã$èfÓ?ä–åyÏcÐçLåyӈé7äÏÉãoêgÓ!ëÑÄÏråyÓ

÷ ÷

æ7î ãÄÏråyîfÒyӈé õ¢çÇæ7ÏÉêgҀî òÄáÄîfãoÓiÏÉãoä€ÓgҀå€Ó?é¢ç î ÑUå%¼æ7æ

çAå€ÏÉæÇÓ ìäÓˆÓIí¶î Ò ÏÉãoäåyçfãoêzÓ¨ïrÔgðƒÕIñUò9Ó?êzÏcó9êgçfôcôÉõ¨å€áÄÓÇòCî äyäÏcö

÷

ÏÉãå€îå€áÄÓ(îfòCÓgãӈãoéÌîfíÊå€áÄÓ/âoÑUå€Ó î$éÄõfÕ0à²áoÓ¢ä€õ$äå€Óˆæ

÷

ÏÉôÉÏrå–õ%îsí¢ç'æ7îfÑÄå€áÒyӈä€îfãoçfãoêzÓËÏÉãÌâoÑÄå€ÓˆäIÏÉäÇçféoéUҀÓ?ä€ä€Óˆé

õ$ÏÉÓgôªéÄä/çòÄҀÓ?ä€ä€ÑÄÒyÓ'çsæ7òÄôÉÏråyÑoéUÓîfíIø&'Ìþ"ç*çsåAå€áÄÓ Ózå

ü

÷ ÷

ÏÉã0òoÑ ôÉÏÉêˆçLå€ÏÉîfã×ï øLðƒÕÇà²áÄϪäÏcã$èfÓ?ä–åyÏcÐçLåyÏcî ãŒÏÉäÐ ÑÄϪéUӈé õ

ä€òCÓgӈé(îfí³Ô æ¦LäˆÕ

åyáÄÓ¢í¶î ҀæËçsãåkä"ÏcãAå€áÄÓiäòCÓgÓ?êká/çfä5å€áÄÓ èfîUêgçfôÄå€Òkçfêzå5í¶îfÒyæËä

÷

à²áoÓ!äòCӈêzå€ÒkçsôCéUÏÉäå€ÒyÏ ÑUå€ÏÉîfã7Ϫä5æ7ӈç äÑoҀÓ?éÇç ä5çí¶ÑÄã9êbå€ÏÉîfã

÷

ç7Òyӈä€îfãoçså€î Ò¢çLå!ç îfÑÄåIÔ´ùUú¢û Õ³à²áÄÓIòÄҀî ò9ӈҝåyÏcÓ?äÊîsíEåyáÄÓ

îfíÄå€áoÓ³æ7î ÑUå€á¼Ò€Ó?äî ãoçLåyîfÒHôÉÓgãoÐså€áHÕ~à²áÄÓ³òCӈçsù èLçfôcÑÄÓ?ävçfҀÓ

Ógå Ñoä€ÏcãoÐAå€áoÓ¼áÄîfåö‚ýÊÏcÒyÓ¼çfãÄÓgæ7î æÇÓgå€Òyõ(ç ä!ý³ÓgôÉôEçfä!þ¯ÿ¡

ü

çfÒ€Ó òÄôÉîsååyӈé/Ïcã/óoÐfÑÄÒyÓiøIçfä³ç í¶ÑÄã9êbå€ÏÉîfã¢îsíåyáÄÓ éÄÏÉäåyçfãoêzÓ

÷

Ϫä²Ïcã$è ӈäå€ÏÉÐ çLåyӈéAçfôcî ãÄÐÇýÊÏråyá¢åyáÄÓ´ä€ò9Ó?êbå€ÒkçsôZéUϪäå€ÒyÏ ÑÄå€ÏÉîfã

îfíòÄϪä–åyîfã7í¶Òyîfæ å€áÄÓ!âoÑÄÓfÕ)(!ãËÏcãå€ÓˆÒ€Ó?ä–åyÏcãoÐ+*Òyӈä€îfãoçfãoêzÓ",

îfí~ä€îfÑÄãoéÕ

Ó.-CÓ?êbå´Ïªä äÓˆÓgã0Ïcã0å€áÄÓËÒyӈéÄÑoêbåyÏcî ãîsí<å€áÄÓ7òÄÒyӈäyä€ÑÄҀÓ7çsæÇö

÷

òoôcÏcå€ÑoéÄÓ'îsíiå€áÄÓ'ä€Óˆêzî ãoé×çsãoéí¶îfÑÄҀå€á×áoçfҀæ7îfãoÏÉêˆä õç

¢¤£¦¥§¢iÛ/Ø©¨ ¢ Ù'Ú ¦ ¢iÚ´Þ¥ ¢Ù'Ý Û¢HÞ @Ú

í{ç êbåyîfÒ0/øAçsã9éŒäî æ7Ó¼éUϪäòCÓgÒkä€Ïcè Ó¼éUӈò9ӈãoéUÓgã9êzÓ¼í¶îfÒ å€áÄÓ

ÎÄî Ò'å€áÄϪä0Ïcã$èfÓ?ä–åyÏcÐçLåyÏcî ã åyáÄÓâoÑÄå€ÓϪäŒî ò9ӈÒyçså€Ó?éçså îUéÄéáoçsÒyæ7îfãÄϪêgäˆÕ*ÿ‚åÇϪäÇù$ãÄîLýÊãÌå€á9çLå7å€áÄÓ¨äå€ÒyÓgãÄÐfå€áîfí

÷

ç Ógå(ä€ò9ӈӈé¤îsíAÔ0ö*æsäAýÊÏcå€á åyáÄÓ'óoãoÐfÓgÒyÏÉãÄÐÌîsí ӈèfӈã áoçsÒyæ7îfãÄϪêgäéÄÓgòCÓgãoéÄävîfãåyáÄÓ¯ôªç ÏcÑÄæî&-Zä€ÓzåˆÕ~ú¢îLý²ö

ü

÷

¤Ëêzî ҀÒyӈä€ò9î ãoéUÏÉãÄТå€îÌÔfÔ (ú¢û'Õ/à²áoÓ ÓzåIϪäí¶îfÒyæ7ӈé ӈèfӈÒ5ç ä–åyҀî ãÄÐè`çfҀϪçLåyÏcî ãIîfí9åyáÄÓ¢Ó©-@ӈêzå5ÏÉä5ãÄîså5î äÓˆÒ€è ӈéZÕ

ü

÷ ÷

õ0ç'âoÑÄÓ Ózå–ý³ÓgӈãêgçsÑÄåyêkáÄî Ñoê7ôcÏÉòoä¼çsãoé*ÏÉãå€ÓgÒkçfêzåyä´çså

÷ ÷

ç éUϪäåyçsã9êzÓîfí!ÄÕ sæÇæ ýÊÏcå€áå€áoÓôÉç ÏÉÑÄæ¨Õ à²áoÓçsÏÉÒ ÿ¨ã'îfÒkéUӈҲå€îËÓ©1ÄçfæÇÏÉãÄÓ´ç7òCî äyäÏ ôÉÓ´êgçfÑoäÓîfí~åyáÄÓ´ä€ÑÄòUö

÷ ÷

Ϫä/Ð ÑÄϪéUӈé¤åyáÄÒyîfÑÄÐ á çÌåyÑ Ó%ýÊÏcå€á çæ7îLè`ç ôÉӌòÄϪä–åyîfã òoҀÓ?ä€ä€Ïcî ãìfä€òoçLåyÏÉçfôoäyêgçsã9äEýÊÏråyá7çáoîsåö‚ýÊÏÉҀÓÊçsãoÓgæ7îfæ7ÓzåyÓgÒ

ÒyÓgòÄÒyӈä€ÓgãåyÏcãÄÐå€áÄÓ*æ7îfÑÄå€áçsãoé èfîUêgçfôiå€Òkçfêbå?ìiä€ÓgÓ%óoÐfö çfҀÓ(åkçsùfӈã çLåAç*ò9îäÏcå€ÏÉîfã¤îfí øÄÕ 0æ7æ éUϪäåyçsã9êzÓ¨í¶Òyîfæ

÷

ÑÄÒyÓ%ÔfÕ¤à²áÄÓ¨æÇî ÑUå€á¤Òyӈä€îfãoçfãoêzÓ¢îUêgêzÑoÒyä7çLåAç î ÑUå'Ô" åyáÄÓAâ9ÑÄÓfÕà²áoÓ ÓzåÇèfÓgôÉîUêzÏcå–õòÄÒyîsó9ôcÓAϪäIç`è ÓgÒkçsРӈé0îLèfÓgÒ ü

Î~ÏÉÐfÑÄÒyÓG9ÕKJ

žW™?“¨’y«{› Ÿ'™ˆ©"”–¦’ ‘?’kžW™L¡k•¤”ƒ /À$’kžr£(§ ’b˜g“¨› œ–’b£¢ºs

µHm·@L"žW’y©c”7“¨•r£ ’?Å+M<™ˆœ–§´˜gž²™?ŸU’yœ˜z”–•W™?1¬x§´˜gÈĸ ‘`’kžW™L¡k•¤”ƒ ONP

§RQg“±SLĜ–•W—?¦L”Ê“¨•r£ ’?Ų°<¡k™ˆ›“‚”–•W¡b˜gž¥E˜b‘`’™? ž¤ '¬x§´˜gÈĸ¢‘`’kžW™L¡k•¤”m

Ni§RQg“±€¸

åyÏcî ã/îfíZå€áÄÓiéÄçLåkçIä€áÄîLý¢ä"åyáoçLå<å€áÄÓièfӈôcîUêzÏcå–õ¼óoÓgôªéAîfíZå€áÄÓ

ç êzî Ñoä–åyÏÉêIý³ç`è ӴϪä ãoӈçsÒyôÉõ(ÏÉãòoáoçfä€ÓIýÊÏcå€áå€áÄÓÇò9îäÏcå€ÏÉîfã

÷ ÷

îfívåyáÄÓ Ózå!ç äÊêgçfã Ó´äÓˆÓgã¨ÏÉã¢ó9ÐfÑÄÒyÓ¤oì îsååyîfæ¨Õ

ü

ÚÇÜ0TU /Û¢ÝVW /ÙX¢¤£¦¥§ Y /ÙZ ڴ؈ÜAÙ Ü![ Ú0\Z¢

HÞK¥§¥ Û¦¢§]H؈ÜËÙ^¢@[H[H¢ ß¼Ú

ÿ¨ã óoÐfÑÄÒyÓO×ç×ä€áÄÏríxåŒîfí´å€áoÓÌèfÓgôÉîUêzÏcå–õ¤ò9Ó?çsù çfãoé ç

Î~ÏÉÐfÑÄÒyÓ!ÄÕ

¹C™?ŸoÅ!·v’kžW™L¡k•¤”m ¨˜g“ ˜I©ª› ¡y”–•W™?¨™g©"”–•W§ ’?¸2Ǖr£f«

÷

ä€ôÉÏcÐ á åyôcõ'ÏÉãoêgҀÓ?çfä€Óˆé'ýÊϪé$å€á%Ϫäî ä€ÓgÒyèfÓ?éZÕ§( æ7ӈҀÓ7äáoÏríxå

£ žW’?Å!°¯§ ŸžW•¤”–›$£ ’ ™ˆ©E”–¦ ’“¨•W—?˜ˆž~™g©5§ •W¡yœ–™?Ÿ ¦™? ’03¢¬x§ ™ˆ› ”–¦

ϪäiãÄîsåiôcÏÉùfӈôcõ(åyî/òÄÒyîUéUÑoêzÓÇçAôªçsÒyÐfÓÇêkáoçsãÄÐ Ó´ÏÉã'å€áÄÓÇáoçfҝö

œ–’k“¨™?˜g”–™gœ€±¢©ª™gœŸ •r“‚”–™ˆŒŸU™?“¨•¤”–•W™?“ ºU’y”ƒ¥~’k’kÌ»54/˜g$£%»76g¡k§7¸

8~™ˆ”¨”–™ˆ§7Å"µ@™ˆ“¨•W”–•W™ˆ/™g©”–¦’"Ƃ’y”i¬x™?ŸU’kË“:9L›˜gœ–’k“±5˜ˆ£Ë˜ˆ¡k™ˆ›“‚«

æ7î ãÄÏÉê0êzî ã åyÓgãå¢îfí å€áÄÓ0èfî ôcÑÄæ7Ó'â9îLý Ïcãå€îå€áÄÓ0âoÑUåyÓfÕ

”–•W¡b˜ˆžÄ¥"˜b‘`’ ¬ªÀžržW’b£¼“:9s›˜gœ–’k“±v¥5•¤”–¦Ëµƒ·¦§ ’b˜ˆ“¨›fœ–’k§ ’kL”b¸

÷

à²áoÓ%è îfҀå€Ó.1 äáÄÓ?éÄéUÏÉãÄФÓgè Ógãå€ÑoçfôcôÉõ¤áoçfä(åyî Ó%åyçfùfӈã

÷

ÏÉãå€î7çfêˆêzîfÑoã å?Õ~à²áoÓiä€æËçsôÉô@ä€áÄÏcíxå³îfíå€áoÓ òÄá9çfä€Ó Ózå–ý³ÓgÓgã

åyáÄÓ¯òCî ä€Ïcå€ÏÉîfãIçfãoéiå€áÄÓ³çfêgîfÑoäå€ÏªêEèfӈôcîUêzÏcå–õ óoÓgôªéÏcã9éUÏÉêˆçLåyӈä

÷

æËçsã$õ7êzõUêzôÉӈä³çsã9éËóÄå€å€Óˆé/ýÊÏråyá¢ç;<Ϫêkù$ôcӈõËéUÏÉäå€ÒyÏ ÑUå€ÏÉîfãHÕ

÷

åyáoçLå~å€áoÓ Ózå~ÏÉävÓgè Ógãå€Ñ9çsôÉôcõéUÒyÏcè Ógã´ÒkçLåyáÄÓgÒ õiå€áoÓ5å€Òkçsã9ä–ö

ü

à²áÄÓ¢êgÓgãå€ÒkçsôoèfÓgôÉîUêzÏcå–õ´Ïªä¯ä€áÄîLýÊãËçfä~í¶ÑÄãoêzå€ÏÉîfãAîfí9åyÏcæ7Ó¢ÏÉã

÷

è ÓgÒkäÓ¢âoîLý×å€áoçfã õÇåyáÄÓ¢åyÒyçfãoäè ÓgÒkäÓ³âoîLý ç êgêzӈôcӈÒyçså€ÏÉîfãÕ

÷

óoÐ ÑÄÒyÓ<oÕ~à²áoÓgÒyӳϪäEåyáÄÓ¢çò9Ó?çsù´îfí@ç î ÑUå¢Ô">= í¶îfÒ¯ò9îäÏcö

åyÏcè ÓÊå€ÏÉæ7ӈä³çfä"åyáÄÓ Ógå<æ7îLèfÓ?ä5éUîLýÊã$ý²çsÒkéÄä"êzÒyî äyäÏÉãÄÐåyáÄÓ

ü

ß7ÜAÙßR <ÞVH؈ÜAÙ

÷ ÷

ôªç ÏÉÑÄæ¨ì<ä€ÓgÓ îså€å€îfæ òoçsҀåËîsí!åyáÄÓ(óoÐfÑoÒ€Ó Õ×à²áÄϪäÇí{çfêzå

à²áÄÓ²æ7îfÑUåyáIÒyӈä€îfãoçfãoêzÓ5í¶îfÒ~âoÑUåyÓ³çfãoé´ÒyӈêgîfÒkéUÓgÒvòÄôÉç`õö

÷

êˆçsã Ó(éUÓ?éUÑoêgӈéçfôÉä€î'í¶Òyîfæ òÄÒyӈäyäÑÄÒyÓËæ7ӈç äÑÄÒyÓgæ7ӈã åkä

÷

ӈÒyä¢ÏÉä!æÇӈãå€ÏÉîfãÄÓ?é¢ÏÉãï Gsðvç ä ӈÏcãoÐAÏcæ7òCîfҀåyçfã åÊí¶î Ò¢òÄÒyîsö

÷

ï Lð‚Õ~à²áoÓÊò9Ó?çsùIϪä¯çfòÄòÄÒyî1UÏcæËçLåyӈé õÇç0?çsÑoäyä€ÏÉçfãÇÏÉãËîfҀö

éÄÑoêzÏÉãÄÐ0ç'å€îfãoÓ¢ëÑoçsôÉÏcå–õ0ýÊáÄϪêkáϪä´ôÉӈäyäIä€áÄÒyÏcôÉômÕÿ¨ãÌå€áÄϪä

éUӈҳå€îAëÑoçsãåyÏrí¶õ/êká9çsãÄРӈä³Ïcã(òCî ä€ÏråyÏcî ã(çfãoé(áÄÓgÏÉÐfáåˆÕ)( ä

ò9çsòCÓgÒAåyáÄÓç ä€ä€Óˆäyäæ7ӈã åAîfí´çsã Ó©-@ӈêzå¢ÏªäËæËçféUÓýÊáÄÏÉêká

÷

ÏÉãoéUϪêgçså€Óˆé(ÏÉã(å€áÄÓóoÐ ÑÄÒyÓ õ(çsÒyҀîLý¢ä³çLåÊåyáÄÓ ò9Ó?çsù¢ò9îäÏcö

ä€ÑÄòÄòoҀÓ?ä€ä€Óˆäå€áÄÓ5ӈèfӈã áoçsÒyæ7îfãÄϪêgävçLåvçÊòoҀî ò9ӈҀôÉõ ä€áoçsòCӈé

÷

åyÏcî ãAå€áoÓiêkáoçfãÄÐfÓ ýÊÏråyá/åyáÄÓòÄϪä–åyîfã/òCî ä€Ïcå€ÏÉîfã/Ϫä³î è$Ïcî Ñoä

è îUêgçsôvå€ÒkçfêzåˆÕ à²áÄÓ¼óoãoéUÏÉãÄТîsí<ç/òÄáoçfä€ÓÇæÇîUéUÑoôÉçså€ÏÉîfãŒîfí

ç ä¢åyáÄÓ¼æ7î ÑUå€á0ҀÓ?äî ãoçsãoêgӴϪä!åyÒyç`è ÓgÒkäÓ?éZÕ@( äiÓ©1Uò9Ó?êbåyӈé

åyáÄÓÇèfÓgôÉîUêzÏcå–õ¢òCӈçfù'çLå!åyáÄÓIæ7î ÑUå€á0ҀÓ?äî ãoçsãoêgÓ æËç`õ¢ÏÉãUö

åyáÄÓIä€áÄÏríxå ÏÉä êzîfÒyÒyÓgôªçLå€Ó?é/ýÊÏcå€á¨åyáÄÓ´òÄÒyӈäyä€ÑÄҀӴçfæ7òÄôcÏcå€Ñ9éUÓ

éÄÏÉêˆçLå€Ó´çfã¢Ó.1UòÄôÉçfãoçLåyÏcî ãÕ

÷

îfíiåyáÄÓæ7î ÑUå€á Òyӈä€îfãoçså€î ÒAçfä/êˆçsã Ó0ä€Ógӈã¤í¶Ò€î æ åyáÄÓ

óoÐ ÑÄÒyÓ¤oÕ

Û¢@[H¢iÛ¦¢iÙߢ

ñ$ÏÉãoêzÓ5éUÓgåyçfÏcôªäZîsí$åyáÄÓ Ózå~çsÒyÓEéÄÏBAAêzÑoôråvå€îÊÓ©1$åyÒyç êbåýÊÏcå€á

ü

#

ú%CD(´ì?æ7ӈçfä€ÑÄÒyÓgæ7ÓgãåkäZýÊÏråyáIþ5ÿ¡ ÿ:E(×ñ$õUä–åyÓgæËä5$çféoéÄä

_7`bac`"d&e5f"gih)e5jk%lY`"m]nogqpirts.f&h:giuvwyx{z7|t}~t€7‚Rƒ „.†  }B‡{ˆ@ƒq‰‹Šy Œ†

ÏÉãUí¶îfÒyæËçLåyÏcî ã îfã´åyáÄÓ³èfÓgôÉîUêzÏcå–õ!ó9ÓgôªéZÕ~à²áoÓ<è ÓgôÉî$êgÏrå–õòCӈçfù



‡ |iŽ¡Œ"‘ †S‡ Ž|:’+“I”•†S–S}—† ˜™ƒi‰0š›Œ{‘cœ{†t„‚Rƒ „©†S |¤l]žFŸc .¡

çfãoéAÏcåyä¢éÄÓgòCÓgãoéUӈãoêzÓî ã/åyáÄÓæÇî ÑUå€á¨Òyӈä€îfãoçså€î Ò¯ôÉÓgãÄÐfå€á

¢‹£

•l+¤¦¥7§` ¨7©"v&ª7«7«¬¡­©"_ ®%¯°7®7®7®©±

°"`b²`´³µ`¶y¥7§ ·i¸¹e5jv•‚0ƒ©Œ{Žx!º† |¡ƒ7‡&€©‡&~i†R†»]†t~ Ž|H}‡¦Žx"†½¼yŒ"Ž¾†¡¿

Ϫä³Ò€ÓˆòÄҀîUéUÑ9êzӈéAÏcã/å€áÄϪä²Ó©1UòCÓgÒyÏcæ7Ógãå?ÕF( äÊäÓˆÓgã¢Ïcã/óoÐfÑÄÒyÓ

²7l) {l+¤¶¥7§` ª5Àv"À_SÁ%¯º_ «©Á5©±

G9ìfÒyÏcÐ á å³äÏªéUÓfìfå€áÄÓiéUîLýÊã$ý³çfÒyéoäEæÇîLè$ÏÉãÄÐ Ózå³òÄÒyî$éÄÑoêzÓ?ä¯ç

ü

Â"`bac`.d&e5fgih]h:·•e5§`ov¦Ã'ƒ7¡Ž¾†¡Ä |qx"†t„©„7}B‡{ˆ¹}B‡¤|tŽ¾†i€.„7zHƒ7|¡~ }B €7Ž}ƒ7‡§ƒq‰

äå€ÒyîfãÄÐ èfî ҝåyÓ©1Zì?å€á$ÑoäEÓ©1UòÄôªçsÏÉãÄÏÉãÄТí¶Ó?çLåyÑÄҀÓ?äví¶îfÑÄãoé¼ÏÉãIåyáÄÓ

¢ ¢‹£

€‹¼yŒ†)ƒ7—ˆ.€7‡<œ&} œ{†‹l] ll]žFŸc yl¤¦¥7§`.¨7°"v"¨7©7¬¡¾¨©Á7Á

÷

ú%CD(´ÕZà²áÄÓËþ5ÿ¡ ä€Ózå€ÑoòϪäiç ôÉÓIåyî(éUÓgå€Ó?êbåiå€áÄÓËçfêgîfÑoäö

¯q_ «©Á5©±

À`ÆÅY`{ {rtsh:rtÇ&v ÈÉ}—†Š¶tʃ©Ž†:¿˶k"n ·ino¥7j¹ac`. .rts¥5·º·SÌ p 'Í¥7s"jh7v"Χe5noj"Ï

åyÏÉê è ÓgôÉîUêzÏcå–õAóoӈôÉéZÕHC Ïcå€áoîfÑUåiîfòCÓgÒkçLå€ÏÉãÄÐ7åyáÄÓ âoÑUåyÓIÏcå!Ϫä

_ «©Á5ª

÷

êgҀÓ?çLå€Ó?é õ*çäòCӈçfùfÓgÒ©`áoîfÒyã*ä€õUä–åyÓgæ êzî ÑÄòÄôÉӈé*åyîŒåyáÄÓ

÷

î ò9ӈã Ógãoé¤îfíiåyáÄӌâ9ÑUå€Ó îUéUõ ÕI(³åAå€áÄÓ%ä€çfæ7Óäî ÑÄãoé

òÄÒyӈäyä€ÑÄҀӨçsæ7òÄôÉÏcå€ÑoéUÓ¨åyáÄÓ¨èfӈôcîUêzÏcå–õÌóoӈôÉé×Ϫä7æ7ӈçfä€ÑÄÒyӈé

÷

çså³åyáÄÓ ôªç ÏÉÑÄæ¨ì9ä€ÓgÓióoÐ ÑÄҀÓG9ìÄҀÏÉÐfáå¢ä€ÏÉéUÓ Õ5à²áÄÓ ÓgèLçfôcÑoçsö Embouchures and end effects in air-jet instruments.

Joe Wolfe and John Smith

School of Physics, University of New South Wales, Sydney NSW 2052, Australia, [email protected]

In the flute family of instruments, the control oscillator is an air jet, which acts at an end of the bore that is open to the atmosphere. The air jet thus experiences a radiation impedance determined by the size of the aperture at the jet and the solid angle it subtends. These are varied by the player's embouchure, the upper lip and face. To adjust the intonation, players of the transverse flute rotate the instrument about its long axis, thereby partly covering the hole, varying the radiation impedance and changing the end effect. In the end-blown shakuhachi, players move the flute and the head in the vertical plane to achieve even greater changes in the end effect and pitch. We report measurements of the radiation impedance of the playerÕs embouchure as 'seen' by the instrument Ôlooking outÕ. Independently, we measured the acoustical impedance of the instrument itself, measured at the position of the jet, but 'looking into' the instrument. Adding these gives the total effective impedance of the instrument in the playing configuration. Changes in this impedance with different embouchure can account for most or all of the observed changes in pitch, and may also contribute to changes in timbre.

INTRODUCTION by changing the shape of 'ears' on either side of the jet, which varies the solid angle for radiation, but The air-jet family of instruments includes side blown recorders do not use this technique. We report here the flutes such as the Western flute, and end-blown flutes results for the shakuhachi and transverse flute. such as organ pipes, recorders and the Japanese shakuhachi, which is gaining renown because of its popularity in 'world music'. Because they are excited by an oscillating air jet, they are open to the atmosphere at the point of excitation. The jet is thus loaded by the acoustic impedance of the bore and that of the radiation field at the hole left open at the jet. A similar situation in the organ pipe has been analysed by Fletcher and Rossing [1], who conclude that the effective impedance acting on the jet is that of the bore in series with the radiation impedance. The radiation impedance or end effect can be varied by varying the size of the hole at the jet, the solid angle available for radiation, or both. In the transverse flute, this is achieved by rotation of the instrument so that the player's lower lip occludes more or less of the embouchure hole, and the upper lip and the rest of the FIGURE 1. The shakuhachi and player Riley Lee. face baffle more or less of the solid angle. Flutists use this technique to play in tune under varying MATERIALS AND METHODS conditions of loudness and required pitch adjustment. In the shakuhachi, the player's chin closes the end The impedance spectra of flutes and shakuhachis, of a nearly cylindrical pipe (Fig 1). A chamfer without players, were measured as described provides both the edge (which usually includes an previously [2,3], except for the treatment of the insert of a hard material) and the embouchure hole. radiation load at the embouchure. To measure the Players vary the position of the chin, the angle of the impedance of the end effects, the impedance head was instrument and the angle of the head to effect large placed inside the embouchure end of a flute or a changes in the area and angle available for radiation. shakuhachi, so that it measured the impedance of the This gives the instrument a great flexibility in pitch external field in series with a small section of known and timbre, which have become important geometry. The field impedance was calculated using components of both traditional and contemporary the transfer matrix. playing styles. The players were distinguished concert and With organ pipes, fine tuning adjustments are made recording artists. They were asked to mime the embouchures required for normal playing, and for sufficient to account to first order for the observed playing to produce notes with different intonations. tuning effect. This procedure is highly reproducible: it is what an experienced musician must usually do just before beginning to play if s/he is to play in tune.

RESULTS AND DISCUSSION

FIGURE 2. The impedance 'seen' by the shakuhachi.

Figure 2 shows the impedance of the shakuhachi player's embouchure as used to play the notes ro and ro meri. Both are played with all finger holes closed, but the former has the normal pitch and the latter is a semitone lower. Note that the impedance increases roughly at 6 dB/octave, but that the meri embouchure has a higher impedance. The resonances of the player's FIGURE 3. Impedance of a shakuhachi plus the vocal tract are visible as small features (at around 2.4 radiation load measured at the embouchure. and 3 kHz in the meri curve, higher in the normal curve). Although the resonances of the tract may be strong, the narrow aperture between the player's lips is ACKNOWLEDGMENTS only a small fraction of the solid angle for radiation. In measurements on a beginning player (one of the We thank Riley Lee, Geoffrey Collins, Tom Deaver, investigators) the vocal tract resonances are much John Tann and the Australian Research Council. more visible. The scatter at low frequency is due to background noise. This was smoothed over ±10 Hz for the calculations yielding the figures below. REFERENCES The sum of the instrument and embouchure impedances are shown in figure 3: the upper figure 1. N.H. Fletcher and T.D. Rossing, The Physics of Musical shows the normal embouchure and the lower the meri Instruments. New York, Springer-Verlag, 1998. embouchure. (Measurements of impedance for the shakuhachi alone were made previously [4].) 2. J. Wolfe, J. Smith, J. Tann, and N.H. Fletcher, J. Sound & The frequencies of the minima are lower in the meri Vibration, 243, 127-144 (2001). case by 140, 100 and 65 cents, and are also less 3. J. Wolfe and J. Smith, "Acoustics of the air-jet family of harmonic than those of the normal embouchure. This instruments", in Proc. 7th Western Pacific Regional Acoustics is sufficient to account for the observed difference (1.0 Conf. Kumamoto, Japan, pp 575-800, 2000. semitone flat) produced in a played note. The meri result also shows shallower impedance minima at 4. Y.Ando and Y.Ohyagi. J. Acoust. Soc. Jpn. 6, pp 89-101 high frequencies. Along with the decreased (1985). harmonicity, these may be in part responsible for the 5. J. Wolfe, J. Smith and J. Tann. "Flute acoustics" darker timbre of the meri notes. www.phys.unsw.edu.au/music/flute. Similar results are obtained for the transverse flute. These are reported in greater detail elsewhere [5]. We conclude that, although the jet speed and other parameters may be adjusted, the end effect is

Experiments on mouth-tones during transients and steady- state oscillations in a flue organ pipe

Fabre B.,Castellengo M.

Laboratoire d’Acoustique Musicale, Univ. Paris 6, 4 place Jussieu, 75252 Paris Cedex 05, FRANCE

Attack transients are known to be very important for the perception of the tone quality of flue organ pipes. This may be related to the complexity of the physical phenomena that can take place during the transient. This complexity turns the physical understanding and modeling to a challenge. Experimental investigation of the attack transient shows phenomena that can be related to an edge-tone oscillation or to a pipe-tone oscillation. Oscillating frequencies in an edge-tone geometrical configuration are compared to those obtained in an organ pipe configuration. Measurements show that the edge-tone like oscillation that can take place at some stage during the attack transient can also appear during steady-state oscillations above the oscillation threshold.

Time stages of the attack transient INTRODUCTION As previously discussed by Fabre [2] and Verge Attack transients in musical sounds are known to [3], the attack transient can be separated in four time be very important for perception. Recent work by stages. Rioux [1] shown that, in the case of flue organ pipes, the attack transient plays a major role in the subjective Stage 1 is the very beginning of the transient. The tone quality. Furthermore, the organ builder gives foot pressure rise is pushing an initial flow from the special attention to the attack transient during the final flue towards the labium acting as a volume injection adjustment of the pipe geometry known as “voicing”. and giving rise to the initial acoustic pulse that triggers The physics involved during transients is complex the whole transient as shown by Verge [3]. From since it mixes unsteady flow dynamics with acoustics. experimental results, theoritical investigations and If some aspects of organ pipe physics are well enough flow simulations, it appears that the duration of stage 1 understood to allow the present models to predict lies between 0,5T and 3T where T is the inverse of the steady-state oscillation in the pipe as far as order of frequency of the first pipe resonance. magnitudes are concerned, it is definitely not the case for attack transients. The purpose of this paper is to Stage 2 starts when the flow reaches the labium. present experimental data that show the different The complex interaction with the labium acts as a phenomena that can be observed during the attack source on the fluid, and may act on the jet itself. An transient and to discuss how those phenomena can be oscillation often takes place which may induce an accounted for by the present models. acoustic resonance (longitudinal and/or transversal resonance of the pipe). The oscillation may as well not induce any acoustic resonance, but instead, rely upon the direct action of the source on the jet like that EXPERIMENTS observed in the case of an edge-tone. Experimental works [2,5] indicate that oscillations during stage 2 appear at frequencies much higher than the Experimental setup fundamental of the future steady-state. Flow visualization [3] indicate that the flow behaviour may Experiments were carried on a metal ogan pipe be very complex at that time. During stage 2, the with circular cross section of 27mm diameter and threshold for auto-oscillation is not reached but the 312mm length. The flue exit-labium distance is 7mm pipe starts to accumulate acoustic energy at the and the flue height is 0.20

During stage 3, the system has reached the the pipe resonances. Despite this possible frequency oscillation threshold : the jet velocity as well as the matching, the systems seems not to be locked on one acoustic energy already stored in the pipe are high of the pipe modes : its frequency appears to still evolve enough to maintain auto-oscillation. The jet independantly of the pipe resonance as seen on figure mouvement is locked on one of the longitudinal modes 1. of the pipe. The transverse jet mouvement is large and the source associated with the jet oscillation is DISCUSSION probably at saturation. The pipe accumulates energy at that frequency so that the acoustic oscillation in the Two different loops are generally considered in pipe is growing for a time depending on the quality lumped model description of flue organ pipes [4]. The th factor Qn of the n pipe resonance[2] : t3~Qnfn. first loop uses a direct hydrodynamic feedback of the Stage 4 is the saturation of the acoustic oscillation in source at the labium on the jet. The second one goes the pipe, which appears [4] to be related to vortex through the pipe resonance. Verge [3] has shown that shedding induced by the acoustic field. During stage 4, the second one is dominant during steady-state the oscillation regime of the pipe may change from a operation. The first seems to be dominant during stage higher to a lower longitudinal mode of the pipe. If this 2 while the system is operated below its oscillation does not occur, stage 4 is very short. threshold. Following Coltman’s experiment, we carried Under normal blowing conditions, the global frequency measurements (fig. 2) on both geometrical duration of the attack transient is dominated by the configurations (with and without pipe) as function of duration of stage 3. However, for some organ pipe the jet velocity. It appears that mouth-tone steady-state ranks like the italian “viola” [5], the special voicing of oscillations can exist below the pipe-tone threshold. the pipe allows to stabilize the oscillation at stage 2 These are characterized by their very low radiation and when mouth-tones and pipe-tones can exist together. their great frequency sensitivity to blowing pressure Apart from this very special voicing, stage 2 may still fluctuations. Further measurements on the oscillation be the most important as far as perception is concerned amplitude should be carried. since it induces an oscillation at a frequency much higher than the steady state oscillation.

Mouth-tones Experimental investigation of the attack transient by Castellengo [5] showed the existence of mouth-tones with edge-tone like frequency behaviour. These mouth tones appear at stage 2 : indeed, several oscillating regimes can co-exist at that time since the system has not reached its saturation. Comparison of the sound produced by the complete pipe and by the mouth only (edge-tone) configuration indicates that the mouth- tone is reinforced when its frequency matches one of

Fig 2 : Steady-state oscillating frequencies on edge- tone system for the first three modes (∆,,+) and on the complete pipe (). REFERENCES

1.Rioux V, Acta Acustica, 86(4) (2000), 634-641. 2.Fabre.B & al, Journal de Physique., Colloque C1 (1992) 67-70 3.Verge M.P. & al., J. Acoust. Soc. Amer., 95(2) (1994) 1119-1132. Acta Acustica Fig 1 Time-Frequency analysis of the attack transient 4.Fabre B. & al, 86(4) (2000), 599-610 Acta Acustica on the edge-tone configuration (left) and on the organ 5.Castellengo M., , 85(3) (1999),387-400 6.Coltman J., J. Acoust. Soc. Amer., 60 (1976), 725-733 pipe configuration (right)

Spontaneous and Induced Spanwise Variability in Self- Excited Air Jet Oscillation

A. Wilson Nolle

The University of Texas at Austin, Austin, Texas 78712 USA

At low Reynolds numbers, a self-excited oscillating planar jet (the edgetone system) approximates two-dimensional flow. Breakdown of edgetone two-dimensionality occurs in ranges of flow velocity where competition between oscillatory modes causes intermittent loss of amplitude, which is not simultaneous along the span. Forced breaking of two-dimensional edgetone symmetry is produced by inclining the edge,. the standoff distance varying linearly along the span. Hot-wire signals at the center of the span contain a spectral triplet. The outer two lines are the main components in the acoustic radiation. Their frequencies closely match those in separate edgetone experiments with the edge parallel to the flue and with standoff distance matching one end or the other of the inclined edge. The hot-wire spectrum at an end of the inclined edge is dominated by just one of these frequencies. The effect of the inclined edge in organ-pipe oscillation, where the edgetone can produce anharmonic partials, is investigated. Examples are found where the anharmonic partials diminish or vanish when the inclined edge is used.

EXAMPLES OF SPANWISE FLOW taneously at the two probes, indicating that the flow VARIATION IN EDGETONE varies along the span. Two modes of oscillation, of frequency ratio about 2.95, are involved. If the same The spontaneous oscillation of a planar jet striking an edge stream velocity is approached slowly from smaller can be regarded as a two-dimensional phenomenon for many values, only the lower mode at 390 Hz is present and purposes. However, flow dependence on the third (spanwise) the probes show similar periodic waveforms. dimension can occur, for example, (1) when the jet has evolved toward chaos; (2) when competing modes of 2000 oscillation interfere intermittently; and (3) when the 1500 edge is not made perpendicular to the stream direction. Horizontal edge This paper will show examples of (2) and (3), and then 1000 report some consequences of (3) in organ pipe oscillation. Reynolds numbers are about 500 to 1500. 500 All results are obtained with a parabolic flow from a 1 0 mm wide channel flue, streamwise length 25 mm, span 38 mm, except that an organ-style jet with sharp 2000 languid will be considered in the final example. The Slanting edge edge or lip struck by the jet is of square-cut sheet 1500 metal, thickness 1.6 mm. 1000

500

0 0 1000 2000 3000 4000 5000 freq., Hz

FIGURE 2. Top, velocity spectrum for horizontal edge; bottom, for slanted edge.

0 10 20 30 40 50 time,millisec Figure 2 illustrates the effect of forced spanwise dependence of the flow. The upper graph shows the velocity spectrum near the edge for a uniform standoff FIGURE 1. Flow in an edgetone apparatus, recorded sim- of 8 mm (flow velocity 16 m/s). When the edge is ultaneously by two probes near the edge, 4.5 mm apart.. inclined so that the standoff varies from 7 to 9 mm Rectangles mark interference epiusodes. Velocity 8.4 m/s. along the span, the spectrum in the lower graph is Flue to edge distance, 8 mm. found. Single lines are each replaced by two strong, well separated components, with a weaker component Figure 1 shows an example of intermittent between. Further work shows that the component of interference episodes.. These do not occur simul- highest frequency is dominant at the low end of the this case, using the inclined edge makes a modest edge, and vice versa. reduction in the anharmonic component at 4.1 times the fundamental frequency, but does not remove it. RELATION TO ORGAN-PIPE

OSCILLATION 0.30

With horizontal lip Edgetone frequencies can occur as anharmonic com- 0.20 ponents in organ-pipe oscillation [1]. Modulation of such components appears to contribute to a buzzing or 0.10 frying sound [2] that is sometimes considered ob- jectionable. The preceding excperiments, showing 0.00 modification of the edgetone by inclination of the edge, 0 1 2 3 4 5 6 7 8 suggest that a slanted lip can modify the edgetone- Frequency relative to fundamental produced anharmonic components in organ-pipe oscillation. Two examples follow. Acoustic spectra in Fig. 3 are from a laboratory organ pipe using a horizontal upper lip, and one using an inclined upper lip (cutup varying from 7 to 9 mm). With the horizontal lip there is a component, also found in edgetone measurements, at 5.25 times the 306 Hz fundamental. With the slanted lip these components disappear (but also amplitude of the second harmonic 4.0 4.5 5.0 in the pipe oscilllation decreases).

1.0 Slanted lip 0.8

0.6 Horizontal lip (microphone) 0.4 0.2 0.0 4.0 4.5 5.0 0 2 4 6 8 10 Frequency relative to fundamental FIGURE 4. Top: Acoustic spectrum of pipe having flue with languid. Passage width 0.29 mm. Ears, projecting 12 1.0 mm, fitted to mouth. Resonator length 661 mm. Flue velocity 15 m/s. Fundamental frequency, 236 Hz. Lower left: 0.8 Detail from upper graph. Lower right: Comparable detail for 0.6 Slanted lip inclined lip (7 to 9 mm), showing reduction of anharmonic (microphone) 0.4 component.

0.2 While anharmonic content can sometimes be reduced 0.0 by inclination of the lip, it appears that the level of 0 2 4 6 8 10 turbulence at the lip can be more important. Also, Frequency relative to fundamental anharmonic behavior is sensitive to pipe scale and to voicing details. FIGURE 3. Organ-pipe simulation, 300 Hz. Open resonator, 38 mm square by 473 mm long. Acoustic spectra with hori- zontal and with slanted lip. 18.3 m/s central flue velocity. REFERENCES

In the final example the flue was the passage 1. Castellengo, M., Acustica-Acta Acustica 85, 387- between the front pipe wall and a languid, as usual in 400 (1999). organ principal pipes. This languid is a machined part, 2. Monette, L. G., The Art of Organ Voicing, New Issues with a sharp edge. Chaotic content in the jet flow was Press, Kalamazoo, Michigan, USA, 1999, pp. 73-74. much greater than before, and voicing adjustments less critical. Anharmonic behavior was difficult to find, but was obtained (Fig. 4) with the resonator lengthened. In Experimental Study of the Velocity Field at the Side Holes and Termination of a Tube D. Rockliffa, J.-P. Dalmontb, D.M. Campbella aThe University of Edinburgh, UK bLaboratoire d’Acoustique de l’Universite du Maine (UMR CNRS 6613), France

The velocity field close to tone holes on a woodwind instrument has a significant effect on the behaviour of the instrument. This is particularly noticeable in the lowest register of an instrument, where acoustical streaming velocities can be quite prominent. Previous investigations have developed theoretical models to describe the acoustical behaviour of side ducts [1], which have been supported by experimental measurements. The non-linear behaviour of tube terminations of varying shapes has also been investigated experimentally and theoretically [2, 3]. This work presents an experimental investigation of the non-linear behaviour at the termination of a tube. Firstly, a purely acoustic measurement of the non-linear radiation impedance is taken. Secondly, Particle Image Velocimetry (PIV) is used to obtain instantaneous full-field maps of the acoustic particle and streaming velocities at the tube termination. The acoustic measurement of the radiation impedance shows a non-linear resistance proportional to the volume velocity at the end of the tube whose value is found from the PIV velocity maps.

INTRODUCTION linear resistive term proportional to the velocity ampli- tude at high intensities. Experiments in the exit of open The acoustic field in the region of a tube termination pipes at acoustic resonance [3] showed that the coefficient has been investigated by many authors. Understanding of proportionality of the resistive term may be affected how the velocity field interacts with the walls at tube by the edge sharpness of the tube termination. Recently, openings is particularly relevant in the study of wood- an experimental investigation of linear and non-linear be- wind instruments, where the behaviour of the acoustic haviour at side holes has been carried out by Dalmont field in the vicinity of the tone holes can significantly in- et al. [4]. A tube with upper end open and a side hole fluence the sound produced by the instrument. In addi- with relative dimensions corresponding to that of a wood- tion, the sound intensity found just inside the tone holes wind instrument was constructed. The tube was excited of a woodwind instrument under playing conditions can at frequencies around the first resonance of the tube by a be very high, leading to non-linear effects such as vor- compression driver excited by a sine wave, and radiation tex shedding. The acoustic flow in the exit of open-ended impedance measurements were taken at the side hole and pipes at resonant frequencies has been previously investi- main tube termination for both small and large amplitude gated [3]. Theoretical models developed by Dubos et al. sound fields. The dimensions of the side holes and degree [1] to describe the acoustical behaviour of side ducts have of edge sharpness was also varied. At low amplitudes, been supported by experimental measurements. The aim of this paper is to examine the non-linear be- haviour at the open end of a cylindrical tube and in a short side duct at sound pressure levels comparable to those found on real woodwind instruments under play- ing conditions. Acoustic streaming and particle veloci- ties are investigated using PIV, and the results are com- pared to acoustic measurements of the non-linear radia- tion impedance using conventional techniques.

Non-linear behaviour at a tube termination

Experimental investigations by Ingard and Ising [2] FIGURE 1. Real part of the non linear part of the shunt showed that the radiation impedance at an orifice is lin- impedance as a function of acoustic Mach number Mh for a side ear for low acoustic field intensities, but develops a non- hole of radius 7mm (taken from [4]). FIGURE 2. A photographic image of the acoustic flow at the FIGURE 3. Acoustic particle velocity map corresponding to open end of tube being excited at a resonance frequency of 985 Figure 2. Hz and sound field intensity of 133.5 dB (re 20µPa).

Results and Discussion the measurements agreed well with theoretical results A photographic image of the instantaneous acoustic [1, 5]. For large amplitude sound fields, the radiation particle velocity field taken at the end of a tube when ex- impedance showed a non-linear part proportional to the cited at a resonance frequency is shown in Figure 2. The velocity in the side hole, as reported in [2]. This was corresponding velocity map at the point in the acoustic found to depend on the inner edge sharpness of the hole. cycle where the particles are travelling into the tube can A graph of the non-linear part of the shunt impedance as be seen in Figure 3. It is clear from this figure that parts a function of acoustic Mach number is shown in Figure 1. of the image did not yield velocity measurements, partic- ularly adjacent to the tube walls, due to flare. Although this is unavoidable when working with cylindrical tubes, it does not affect observations of particle displacements on the central axis of the tube or in the region around the The measurement of non-linear behaviour exit of the tube, which are important in the observation in acoustic sound fields using PIV and measurement of non-linear behaviour of the acoustic field at this point. Experimental work is currently being Particle Image Velocimetry (PIV) is a non-intrusive undertaken, and results will be presented at the confer- optical technique which allows the simultaneous mea- ence. surement of flow velocities at many points in a two- dimensional plane. In a typical experimental set-up, tracer particles suspended in a fluid are illuminated twice ACKNOWLEDGMENTS in a short time interval by a thin light sheet projected through the flow. Light scattered by the particles is The financial support of EPSRC is gratefully acknowl- recorded on two separate frames by a digital camera edged. We thank J.-P. Dalmont for discussions on acous- placed perpendicular to the flow. By determining the par- tic fields at tube terminations. ticle displacement between frames and knowing the time separation, the fluid velocity motion can be calculated. The aim of the work is to investigate the flow field near REFERENCES tube terminations under conditions comparable to those found for a woodwind instrument under normal playing 1. V.Dubos, D.Keefe, J.Kergomard, J.-P.Dalmont, A.Khettabi and C.J.Nederveen , Acustica, 85, 153-169 (1999). conditions. Sound pressure intensities exceeding 160 dB were recorded just inside the exit of a side hole on a clar- 2. U. Ingard and H. Ising, J. Acoust. Soc. Am. 42, 6-17 (1967) inet when played fortissimo. Hence, acoustic particle and 3. J.H.M.Disselhorst and L.Van Wijngaarden, J. Fluid. Mech. streaming velocities are measured at frequencies around 99(2), 293-319 (1980). the first resonance of the tube i.e. 200-600 Hz for sound 4. J.-P. Dalmont, C.J. Nederveen, V.Dubos, S. Ollivier, V. field intensities up to and above 160 dB. Once the vector Meserette and E.te Sligte, accepted for publication in Acus- maps have been obtained, further analysis is completed tica. to quantify the radiation impedance and hence the ampli- 5. C.J.Nederveen, J.K.M.Jansen and R.R.van Hassel, Acus- tude of the non-linear term. tica, 84, 957-966 (1998). Time-domain Simulation of Sound Production of the Sho T. Hikichi and N. Osaka NTT Communication Science Laboratories 3-1, Morinosato Wakamiya, Atsugi, Kanagawa 243-0198, Japan

This paper proposes a physical model of the sho, one of the Japanese traditional musical instruments, with intention of applying the model to sound synthesis. A time-domain simulation clarified the effects of the tube length and blowing pressure on the sound frequency and other attributes. To confirm the model, reed vibration and the pressure variation inside the tube were also measured by artificially blowing the sho. The agreement between the experimental results and simulation is acceptably good for the relationships between the tube length and threshold pressure, and between the tube length and the sounding frequency.

INTRODUCTION

The sho is a free-reed mouth organ that is used in tradi- tional Japanese court music called “Gagaku.” It is mainly composed of a mouthpiece, cavity, and seventeen bamboo pipes. Bronze reeds are glued with resin to the lower side of the bamboo pipes. When a player blows in through the mouthpiece and closes the small finger holes on the tubes, oscillation commences and the reeds start sound- FIGURE 1. Overview of the sho, the Japanese free-reed mouth ing. Some tubes have one slot besides the finger hole, organ. which determine the effective lengths. It is categorized as a free-reed instrument like the harmonica, accordion, and reed-pipe organ. However, unlike the other Western free reed instruments, the reeds of the sho are approximately extension is easy. We focus on the “Ichi” tube (B4, symmetrical, so that the same reed vibrates on both blow- 482.7Hz), because this tube does not have any slot on the ing and drawing. side wall, therefore it can be regarded as a cylinder. The instrument is said to originate in the 3rd or 4th Assume pressure at the upstream of the reed (i.e., pres- ( ) century in China, and there are other musical instruments sure inside the cavity) p t , pressure at the downstream ( ) that work on a similar mechanism: the Chinese sheng, (pressure inside the pipe) p2 t . When the reed is as- the Laotian khaen and others, mostly found in eastern sumed to vibrate at the normal mode, the equation of mo- and southern Asia. For khaen, Cottingham has examined tion of the reed is its acoustical properties and reported that sounding fre- d2x ω dx 1.5WL + r + ω2(x − x )= (p(t) − p (t)), (1) quency decreased with increasing blowing pressure, and dt2 Q dt r 0 m 2 that the frequency changed with changing tube length [1]. The main difference between the sho and the khaen lies where x is the displacement at the tip of the reed, Q the in the position of the reed. In the sho, the reed is located resonance Q value, ωr the angular frequency, and x0 the at the one end of the pipe, whereas in the khaen the reed initial displacement. W , L, and m are the width, length, is at a position L/4 from the end. and mass of the reed, respectively. The derivation of Eq. This paper proposes a physical model of the sho, and (1) is shown in the Appendix in the reference [2]. presents results of a time-domain simulation that was car- From Bernoulli’s equation, the relationships among ried out to investigate effects of the tube length and blow- p(t), p2(t), and volume velocity through the slit U(t) are ing pressure on the sound frequency and sounds spectra. as follows: To model the reed vibration and flow through the reed,

 ρ U(t) 2 ∂ ρU(t)δ we adapt the formulation used in the reference [2]. p(t)=p (t)+ + , (2) 2 2 CF(x) ∂t CF(x) where C is the flow contraction coefficient, which repre- PHYSICAL MODEL sents the effect of the slit configuration. The area of the slit F(x) is described by The physical model of the sho is briefly described. 2 2 1 2 2 1 Here, only one tube is considered for simplicity, but F(x)=W[x + b ] 2 + 2L[a(x) + b ] 2 , where a(x) is the average displacement of the sides of the high frequency components of their spectra increase with reed, and b is the gap width. For a sho reed, we assume increasing blowing pressure. Further, from the frequency x0 = 0. Considering it displaces both ways, and from the relationships among fp, fs, and fr, it was concluded that form of the mode function, a(x)=0.4|x| is derived. the reeds of the sho act as “outward-striking valves.” Since the pipe shape is approximated as a cylinder, we use a simple reflection function of Gaussian type of the form ACKNOWLEDGMENTS  −aexp{−b(t − τ)2}, t ≥ 0 r(t)= This research was supported in part by the Center 0, t < 0 for Integrated Acoustic Information Research (CIAIR), and adapt Schumacher’s method to calculate the pressure Nagoya University. One of the authors (T. H.) is grateful at the entrance of the pipe [3]. Using this reflection func- to Professor F. Itakura, who is leading the center, for his tion, we calculated the pressure inside the tube p2(t) as valuable suggestions.

p2(t)=Z0Uin(t)+r(t) ∗ (p2(t)+Z0Uin(t)), (3) REFERENCES Uin(t)=U(t)+0.4WLdx/dt,

where Uin(t) is the net volume velocity input to the tube. 1. Cottingham, J.P., Fetzer, C.A., Proc. of the ISMA, Leaven- Differential eqs. (1)∼(3) were discretized, and a sim- worth, 1998, pp.261-266. ulation was done in 48-kHz sampling. Given p(t), the 2. Tarnopolsky, A.Z., Fletcher, N.H., and Lai, J.C.S., J. following three variables were calculated: displacement Acoust. Soc. Amer., 108(1), 400-406 (2000). of the reed x(t), pressure inside the tube p2(t), and vol- 3. Schumacher, R.T., Acustica, 48(2), 71-85 (1981). ( ) ume velocity U t . 4. Hikichi, T., and Osaka, N., Proc. of the ISMA, Perugia, 2001. SIMULATION AND EXPERIMENTAL RESULTS 700 blow draw Basic acoustical properties were examined both by 650 simulation and experiment. The properties that were ex- 600 [Hz] amined were: 550s 500 ¥ pipe resonance frequency f vs. threshold behavior, p 450 ¥ pipe frequency fp vs. sounding frequency fs, Sounding frequency f 400 ¥ blowing pressure vs. sounding frequency fs. 350

Because of the lack of space, only one example is 300 100 200 300 400 500 600 Pipe resonance frequency f [Hz] shown here. Fig. 2 shows the measured and Fig. 3 the p f simulated sounding frequency s as a function of the pipe FIGURE 2. Experimental result of the sounding frequency as frequency f . The pitch f shows strong dependency on p s a function of tube resonance frequency fp (’blow’: positive, fp, and the results show similar tendencies. Oscillation ’draw’: negative pressures). commenced only when fp < fs and fr < fs hold, where 700 fr is the reed resonance frequency. A more detailed dis- simulation (Q=30) simulation (Q=400) cussion will be presented in [4]. 650

600 [Hz] 550s

SUMMARY 500

450

A time-domain simulation was done to examine ef- Sounding frequency f 400 fects of the tube length and blowing pressure on sound 350 attributes. Simulation and experimental results show that 300 100 200 300 400 500 600 the model reproduces basic characteristics which were Pipe resonance frequency f [Hz] p observed in the actual instrument. Recorded and sim- ulated sounds have a common feature in the sense that FIGURE 3. Simulation result of the sounding frequency. Investigation of Perceptual and Articulatory Correlates of Tonal Ideals in German and Italian Schools of Classical Singing

K. Verdolinia, B. Storyb, and M. Taylorc

aCommunication Sciences and Disorders, School of Health and Rehabilitation Sciences, University of Pittsburgh, 4033 Forbes Tower, Pittsburgh, Pennsylvania 15260 USA ( [email protected]) bDepartment of Speech and Hearing Sciences, University of Arizona, P.O. Box 210071, Tuscon, Arizona 86721 USA ([email protected]) c245 Atkins Avenue, Lancaster, Pennsylvania 17603USA ([email protected])

We examined perceptual and articulatory correlates of tonal ideals in German versus Italian approaches to classical singing. Sung phrases were synthesized on /a/. Labial, oral, and pharyngeal areas were successively varied relative to a neutral mode, holding glottal source constant. Phrases were rated perceptually by expert, uninformed judges. Primary results were: (1) Both German and Italian ideals were most frequently described perceptually as “front,” “open,” and “bright;” (2) phrases identified as Germanic tended to be produced with a relatively larger pharynx and narrower lips, as compared to Italian phrases, which were most often associated with a slight-moderately open vocal tract.

According to one or more prominent standard sung phrase from “The Heavens are approaches to vocal pedagogy, anecdotally, the Telling” by interpolating the fundamental Germanic tonal ideal in classical singing is frequency parameter through the musical notes perceptually “back,” “open,” and “dark” (or of the phrase, on /a/. All other source parameters “covered”). In comparison, the Italian ideal is were held constant. more characteristically “front,” “closed,” and The vocal tract shape was represented as a 44- “bright” [1]. Various articulatory gestures can section area function. Sections (areas) 1-9 were be conceived—and have been proposed—as considered to be the epilarynx, 10-22 pharynx, physiologically causal to the respective 23-39 oral cavity, and 40-44 lips. Each section perceptual targets [1]. The purposes of the assumes a length of 0.396 cm, producing a total present study were (a) to provide experimental vocal tract length of 17.42 cm (44 x 0.396). The data about salient perceptual attributes of baseline (neutral) area function was for an adult German versus Italian tonal ideals in classical male vowel /a/, based on Magnetic Resonance singing, and (b) to provide quantitative Imaging data by Story [5] (Figure 1). For the information about possible articulatory present study, this area function was manipulated contributions to the perceptual ideals [2]. by independently varying the areas of pharyngeal, oral, and lip regions 100%, METHODS 75%,50%,25%,-25%, -50%, and –75% relative to the neutral configuration, while holding the A series of sung phrases was synthesized using remaining regions neutral. A total of 3 (regions) an approach based on exciting a tubular x 7 (area manipulations) phrases were created for representation of the vocal tract (in the form of presentation to judges. an area function) with a voice source waveform. Synthesized phrases were played to 4 expert The voice source waveform was generated with a judges, 1 female and 3 males, ages 39-59 yr, all parameterized glottal flow model [3]. A digital with normal hearing bilaterally at 20 dB HL to waveguide was used to simulate acoustic wave 8000 Hz, who had taught classical singing for propagation in the vocal tract (energy losses due 17-20 yr. All judges indicated familiarity with to yielding walls, viscosity, and acoustic German and Italian tonal ideals. Judges were radiation were included). The particular asked to use their own internal criteria to indicate implementation as used in this study is called the degree to which each phrase conformed to SpeechMaker [4]. the German and to the Italian ideal (poor, A glottal flow waveform, used for all medium, or good conformation). For each conditions in the experiment, was created for a phrase, judges further indicated, forced choice, Table 1. Percent of tokens judged as Germanic versus Italian, for lip changes.
lip area German Italian +100% 64% 45% + 75% 67% 67% + 50% 75% 83% + 25% 100% 75% - 25% 92% 75% - 50% 67% 50% - 75% 67% 17%

Table 2. Percent of tokens judged as German versus Italian ideals, for oral area changes (
).
oral area German Italian Figure1. Baseline (neutral) vocal tract. +100% 50% 42% + 75% 67% 42% + 50% 75% 92% whether the sound was dark or bright, closed or + 25% 75% 92% open, and back versus front. Each phrase was - 25% 67% 58% presented independently to each judge 3 times, in - 50% 67% 50% random (and different) orders across judges. - 75% 33% 25% Rating sheets were varied to control for possible right/left response biases. A neutral aural Table 3. Percent of tokens judged as German “calibration” phrase was presented prior to the versus Italian ideals, for phx area changes (
). start of each rating session, and after each set of
phx area German Italian five phrases, as a perceptual anchor. +100% 17% 17% + 75% 33% 58% RESULTS AND DISCUSSION + 50% 83% 67% + 25% 67% 75% Perceptual data indicated that the most - 25% 75% 50% common perceptual attributes for tokens judged - 50% 45% 27% 42% 8% as medium or good exemplars of both German - 75% and Italian sounds were “front,” “open,” and “bright” (data not shown). This finding partially ACKNOWLEDGEMENTS contrasts with anecdotal descriptions in the literature of Germanic singing as “back, open, Work supported by NIDCD K08 DC00139. and dark” (or “covered”), and Italian singing as “front, closed, and bright” [1]. Although the REFERENCES percepts of “back,” “open,” and “dark” were uncommon, when they occurred they were most 1. Miller, R., English, French, German and Italian likely considered Germanic and were produced Techniques of Singing, Scarecrow Press, Metuchen, by area reductions in all vocal tract regions. N.J. 1977. Results specific to articulatory manipulations are shown in Tables 1-3. These tables show the 2. Taylor, M.H., Influences of Vocal Tract Shape on percentage of ratings for which phrases were Tonal Quality in the German and Italian Schools of Singing, M.A. Thesis, University of Iowa, Iowa City, judged as “medium” or “good” exemplars of the 1996. German versus Italian tonal ideals. In summary (see Tables), tokens judged as 3. Titze, I.R., Mapes, S., and Story, B.H., J. Acoust. medium or good exemplars of the German ideal Soc. Am., 95, 1133-1142 (1994). were most often associated with narrower lip area and larger pharyngeal (phx) area than tokens 4. Story, B. H., Physiologically-Based Speech rated as medium/good Italian exemplars. Tokens Simulation with an Enhanced Wave-Reflection Model rated as medium/good Italian sounds were more of the Vocal Tract, Ph.D. Dissertation, University of frequently associated with expanded oral cavity, Iowa, Iowa City, 1995. as compared to tokens rated as medium/good 5. Story, B.H., Titze, I.R., and Hoffman, E.A., J. German exemplars. Acoust. Soc. Am., 100, 537-554 (1996). Acoustic Evaluation of the Reconstruction of Heinrich Mundt Pipe Organs in Prague V. Syrový, Z. Otčenášek, J. Štěpánek

Sound Studio of the Faculty of Music, Academy of Performing Arts Prague, Malostranské nám. 13, 11800 Praha 1, Czech Republic, e-mail: [email protected]

The Baroque organ in the Church of Our Lady before Tyn in Prague (1670–73) was reconstructed in 1998–2000. Acoustic measurement for documentation purposes was carried out before and after the reconstruction. The diagnostic method developed enables a detailed study of the plenum and its contributing stops. Results revealed that levels and contributions of lower plenum harmonics were preserved and that higher harmonics were strengthened owing to correction of Mixtur stops.

INTRODUCTION starting transients of C-tones of all stops, one in the position of triad measurement and one placed close to The organ in the Church of Our Lady before Tyn in the organist. The impulse responses measured by the Prague, built by Heinrich Mundt in 1670–73, is among MLSSA measurement system [3] are used to calculate the most famous Baroque organs in Central Europe. frequency dependence of reverberation time (Figure 1). The organ was reconstructed as part of long-term It is possible to separate the harmonics of individual th general reconstruction of the church in 1998–2000. Its tones in each triad spectrum until the 6 harmonic and original specification was preserved together with calculate their levels. These values, established for the original pipes, as well as wind-chests and action. The whole range of the stop, enabled a detailed view of the goal of the reconstruction was to preserve the character properties of the spectrum both in different of instrument's original sound. measurements (Figure 2) and among different stops. Two acoustic measurements for documentation The sound character of the instrument is most purposes were carried out to enable comparison of the expressive in the plenum. The Mundt Great Organ instrument's sound on July 1992 before the plenum consists of octave and quint principal stops; reconstruction, and on September 2000 after the thus in its spectrum the following harmonics dominate: reconstruction. The documentation method used was 1, 2, 3, 4, 5, 6, 8, 10, 12, 16, 20, 24, 32, 40, 48, 64, 80, developed in 1991-92 [1] especially for documenting 96. Each of their frequencies fall into different third- acoustic properties of rare historical organs, including octave bands if the band boundaries change according measurement and diagnostic techniques. to fundamental frequency [4]. The levels of harmonics calculated in the whole instrument range for every plenum constituting stop were used for the assessment METHOD AND RESULTS of the contribution of these stops to the individual harmonic of the plenum, and thus enable examination The basis of the method is digital recording of the of the intoner voicing intentions. One can gain a more sound from all pipes of documented instruments and general view of sound character by averaging levels room acoustic measurement [1]. The quasi-stationary over the octave. Figure 3 gives an example of the parts of the tones are recorded by three microphones results of the diagnostic procedure for averaged values placed in the typical listening position in a church, 4 m in octaves C3 and C4 of the Great Organ. above the floor and with a 2-m span among them. Three neighbouring semitones (triads) are played simultaneously [2] and a mean amplitude spectrum CONCLUSION along with the time signal of the first microphone for each triad is recorded. In addition to all pipe stops, a The results reveal that some reconstructed pipe stops plenum sound of every organ machine is documented. have better balanced levels of harmonics in the whole The sampling rate adapts fluently to the stop foot instrument range. Levels of lower harmonics of the length, as well as to the fundamental frequency in plenum remained practically unchanged, including the cases of stops without repetition (sliding sample rate). size of the contribution of individual stops. The This assures the same discrimination of tones in the strengthening of plenum levels of higher harmonics is spectrum of all triads. Two microphones measure the related to the correction of Mixtur and Cembalo pipes. T9 [s]

8 1992 L [dB] L [dB] L [dB] L [dB] Reverberation Time 2000 H1 H2 H3 H4 80 80 80 80 7 60 60 60 60

40 40 40 40 6 20 20 20 after 20 after after after 0 0 0 0 5 a c ab d g a a b L [dB] L [dB] L [dB] L [dB] H1 H2 H3 H4 4 80 80 80 80

60 60 60 60

3 40 40 40 40 20 20 20 before 20 before 2 before before 8' a Principal b Octava 4' c Quint 2 2/3' 2' d Superoctava 1 1/3' e Quint 1' f Sedecima 1' VI g Mixture 2/3' IV h Cembalo 0 0 0 0 a c ab d g ab a 1

0 L [dB] L [dB] H6 L [dB] H8 0.1 0.13 0.16 0.2 0.25 0.32 0.4 0.5 0.63 0.8 1 1.25 1.6 2 2.5 3.15 4 5 6.3 8 10 80 O3 Octave after 80 f [kHz] 60 80 before 60

40 40 20 after 60 20 after 0 0 a b c e g h 40 b d f g h FIGURE 1. Reverberation time in the Church of Our L [dB] L [dB] H6 H8 80 20 80 Lady before Tyn measured before (1992) and after the 60 60

40 0 40 1234568101216202432404864 harmonic [-] church's reconstruction (2000). 20 before 20 before f [Hz] 0 200 500 1000 2000 4000 8000 0 a b c e g h b d f g h

L [dB] L [dB] L [dB] L [dB] H20 80 H12 80 H16 80 80 H24

60 60 60 60

ACKNOWLEDGMENT 40 40 40 40 20 after 20 after 20 after 20 after 0 0 0 0 b c de g h d f g h d g h d e f g h L [dB] L [dB] L [dB] L [dB] H12 H16 H20 H24 The investigation was supported by the Ministry of 80 80 80 80 60 60 60 60

Education and Youth (Project No. 511100001). 40 40 40 40 20 before 20 before 20 before 20 before 0 0 0 0 b c de g h d f g h d g h def g h

L [dB] L [dB] L [dB] L [dB] H32 H40 H48 H64 80 80 80 80 REFERENCES 60 60 60 60

40 40 40 40 20 after 20 after 20 after 20 after 0 0 0 0 1. Štěpánek, J., Otčenášek, Z., Syrový, V. (1994): f g h f g h f g h g h L [dB] L [dB] L [dB] L [dB] H48 80 H32 80 H40 80 80 H64 Acoustic documentation of church organs, 60 60 60 60

40 40 40 40 Proceedings of SMAC 93, Stockholm, 516-519. 20 before 20 before 20 before 20 before 0 0 0 0 2. Lottermoser, W., Meyer, J. (1966): Orgelakustik in f g h f g h f g h g h Einzeldarstellungen, Verlag Das Musikinstrument.

L [dB] L [dB] L [dB] L [dB] L [dB] H1 H4 H6 3. Rife, D. (1987-90): MLSSA Reference Manual, 80 80 H2 80 H3 80 80 60 60 60 60 60

Version 6.0, DRA Laboratories. 40 40 40 40 40 20 20 20 20 after 20 after after after after 0 0 0 4. Štěpánek, J., Otčenášek, Z. (1995): Comparison of 0 a b g 0 ab g ab dhg a b c e g h a L [dB] L [dB] L [dB] L [dB] L [dB] H1 H4 H6 pipe organ plenum sounds, Proceedings of ISMA 80 80 H2 80 H3 80 80 60 60 60 60 60

95, Dourdan, 86-92. 40 40 40 40 40 20 20 20 20 before 20 before before before before 0 0 0 0 0 a b g a b g a b dgh a b c e g h a before the reconstruction L [dB] L [dB] 80 H8 L [dB] H12 80 after 80 L [dB] O4 Octave 60 80 before 60

70 40 40 20 after 60 20 after 0 0 b d g h b c de g h 60 f 40 L [dB] L [dB] H8 H12 80 20 80 50 60 60

40 0 40 1234568101216202432 20 before harmonic [-] 20 before 400500 1000 2000 4000 8000 40 0 f [Hz] 0 L1 b d f g h b c d e g h L3 30 L [dB] L [dB] L [dB] L [dB] H16 H20 H24 H32 C E G B c d e fis gis b c1 d1 e1 fis1 gis1 b1 c2 d2 e2 fis2 gis2 b2 c3 80 80 80 80 after the reconstruction D F A H cis dis f g a h cis1 dis1 f1 g1 a1 h1 cis2 dis2 f2 g2 a2 h2 60 60 60 60

80 40 40 after 40 after 40 L [db] 20 after 20 20 20 after 0 0 0 0 70 d f g h d g h d e f g h f g h L [dB] L [dB] L [dB] L [dB] H20 H32 80 H16 80 80 H24 80

60 60 60 60 60

40 40 before 40 before 40 20 20 20 20 before 50 before 0 0 0 0 d f g h d g h d e f g h f g h

40 L1 L3 30 FIGURE 3. Mean levels of harmonics of Great organ C2 E2 G2 Bb2 C3 D3 E3 F#3 G#3 Bb3 C4 D4 E4 F#4 G#4 Bb4 C5 D5 E5 F#5 G#5 Bb5 C6 D2 F2 A2 B2 C#3 D#3 F3 G3 A3 B3 C#4 D#4 F4 G4 A4 B4 C#5 D#5 F5 G5 A5 B5 plenum and its constituting stops for octaves C3 and C4 before and after the reconstruction. FIGURE 2. Levels of the first (L1) and third (L3) harmonic of Bourdonflauta 16' closed pipe stop with least square fit, before and after the reconstruction.