ACTA ACUSTICA UNITED WITH ACUSTICA Vol. 104 (2018) 825 –829 DOI 10.3813/AAA.919232

Amplitude Modulation MayBeConfused with Infrasound

Torsten Marquardt1),Carlos Jurado2) 1) UCL Ear Institute, University College London, 332 Gray’sInn Road, London, WC1X8EE, United Kingdom. [email protected] 2) Escuela de Ingeniería en Sonido yAcústica, Universidad de Las Américas, Avenue Granados and Colimes, Quito EC170125, Ecuador

Summary Environmental infrasound is usually accompanied by low-frequency(LF)sounds. Considering that inner hair cell transduction equals half-waverectification, activity of low-frequencyauditory nervefibres may be indistin- guishable whether elicited by LF sound that is amplitude-modulated at an infrasonic rate, or LF sound that is superimposed onto infrasound that “biases” the basilar membrane position. We tested whether listeners are able to distinguish a63-Hz carrier tone, amplitude modulated at 8Hz, from a63-Hz pure tone that wasperceptu- ally loudness-modulated by an 8-Hz biasing tone. Using amaximum-likelihood procedure, 12 participants first adjusted the intensity of the 8-Hz tone so that the perceivedmodulation of the pure tone matched areference amplitude-modulated tone. Both stimuli types were then presented in random order,and participants had to iden- tify presentations which contained the infrasound tone. About half the participants performed close to chance. The best had 81% correct. Experiments with a125-Hz carrier tone gave similar results. Although performance may improve in a2-interval discrimination task, this would not be representative of real listening conditions. Results suggest that slowly amplitude-modulated LF sounds may underlie complaints about environmental infrasound, where measured infrasound levels are well belowsensation threshold. ©2018 The Author(s).Published by S. Hirzel Verlag · EAA. This is an open access article under the terms of the Creative Commons Attribution (CCBY4.0)license (https://creativecommons.org/licenses/by/4.0/). PACS no. 43.66.-x

1. Introduction stimulus types under consideration. Because the inner hair cell releases neuro transmitter only during basilar mem- Noise spectra assessed in response to complaints about en- brane (BM) movement towards scala vestibuli, signals be- vironmental infrasound most often reveal that the sound come effectively half-waverectified. In order to illustrate pressures of spectral components in the infrasound range this operation, the lower parts of the signals that are not (< 20 Hz)are well belowsensation threshold and there- coded by the AN are greyed out in the figure. Comparison fore should be inaudible to human listeners (e.g. [1]). of the remaining upper parts shows schematically that the Commonly,such finding closes the complaint case. The spiking probabilities of the AN fibres in response to those measured spectra, however, often cross the auditory sen- twostimuli are almost identical. sation threshold somewhere between 20 Hz and 100 Hz. Of course, this scenario is givenonly if the stimulus These supra-threshold low-frequencynoise components components are not spectrally resolved by the cochlea. might have pronounced envelope fluctuations with spectral This might easily be the case at its very apical end be- content well below20Hz, which might be easily mistaken cause the lowest auditory filters have relatively wide spec- as containing supra-threshold infrasound. tral tuning and more importantly,the lowest stimulus fre- Using simple sinusoidal stimuli, we tested under lab- quencythat has acharacteristic place on the human BM is oratory conditions whether listeners can actually distin- assumed to be as high as 50 Hz (see reviewing remarks in guish between stimuli mixes that contain true infrasound [2, pages 51–52]), if not even 80 Hz [3]. In other words, all and stimuli that are amplitude-modulated (AM) at an in- frequencycomponents lower than this share the very api- frasonic rate, i.e. do not actually contain frequencycom- cal end of the BM as their characteristic place (the place of ponents below20Hz. This might well not be the case if maximum vibration amplitude). The condition illustrated one considers the similarity in auditory nerve(AN)output in Figure 1istherefore not unlikely,and our test results that these twostimuli produce. Figure 1illustrates the two showthat listeners have indeed great difficulty in distin- guishing these twotypes of stimuli with carrier tones of Received9March 2018, 63 Hz and even 125 Hz. accepted 9September 2018.

©2018 The Author(s). Published by S. Hirzel Verlag · EAA. This is an open access article under the terms of the CC BY 4.0 license. 825 ACTA ACUSTICA UNITED WITH ACUSTICA Marquardt, Jurado: Amplitude modulation confusion Vol. 104 (2018)

2. Methods 2.1. Instrumentation The infrasound source used in this study is identical to that used by Kuehler et al. [4]: From ahermetically en- closed electrodynamic loudspeaker,an8-Hz biasing tone wastransmitted via an 8-m long polyethylene tube to the listener’sear.(It wasoriginally developed to be used in MEG and fMRI experiments.)For the last 40 cm, the 14 mm inner diameter of this tube wasreduced via acoupling into amore flexible tube with 2.5 mm inner diameter that took at its end an audiometric ear tip (ER3-14A, Etymotic Research)that washermetically fitted in the listener’sear canal. Before, in-between and after all psychometric mea- surements, the fitting wastested by measuring the SPL in situ with aminiature microphone (Knowles FG-23453) Figure 1. The twostimulus types considered in this study: abi- that wascoupled to the ear canal via asmall plastic tube ased tone (upper panel)and an AM tone (lower panel). The half- (20mmlength, 1mminner diameter)that penetrated the wave rectified versions of the signals (only the upper parts)are foam of the ear tip. It wascalibrated in a1.3-cm3 cavity almost indistinguishable. using aBrüel &Kjær 4153 microphone. Asecond such plastic tube penetrated the ear tip foam to deliverthe other sounds into the ear canal. These were produced by asmall biased were set to equal peak level. Levels corresponded insert earphone (ER4B, Etymotic Research)that wasdi- to either 40 phon, or 50 phon of the pure tone (ISO 226). rectly drivenbythe line output of the audio device (RME The phase of the amplitude modulation and that of the 8- Fireface UC). Using separate sound sources ensured that Hz biasing tone wasindependent and random in each trial. AM due to hardware non-linearities were <1%. In con- The duration of all stimuli was1200 ms, including 250 ms trast, the line output of the 8-Hz biasing tone waslow-pass cosine ramps at onset and offset. filtered (6 dB/octave,passive,fc=10 Hz)before power amplification (BEAK Type BAA120). 2.4. Procedures This filter,together with the acoustic low-pass filtering effect of the 8-m polyethylene tube and appropriate elec- The measurement of each subject, including breaks, took tric attenuation, made sure that at maximum electric out- up to three hours. It wasdone in the ear of their pref- put of the audio device, the sound pressure in the ear canal erence. Before the discrimination task could commence, could not exceed 105 phon [5, 6]. LRBT had to be determined for the eight AM reference stimuli. Initially,subjects made themselves familiar with 2.2. Subjects the infrasound biasing tone and its effect on the pure tone in asimple LRBT adjustment procedure (buttons “up” and Tenfemale and four male normal-hearing subjects (18to “down”), starting from lowlevels. Forall eight condi- 49 years), without self-reported hearing abnormality,were tions, theywere asked to set the biasing tone levelsothat recruited. Their auditory threshold for 8Hz, 63 Hz and the pure tone in the second interval wasperceivedwith 125 Hz wastested using the standard audiometric proce- equal modulation as the leading reference AM tone. These dure (but with 3-dB instead of 5-dB steps). None of the served as individual starting levels for the subsequent, subjects had thresholds above 10 dB HL [5, 6]. However, more accurate maximum-likelihood-tracking (MLT)pro- one male and one female subject were unable to perform cedure. Here, both kind of stimuli were presented in ran- the psychometric procedure so that only 12 subjects com- dom order in twointervals. In a2AFC task, the subjects pleted the study. were asked: “Which of the twointervals wasstronger modulated?” The tracks of the eight conditions, each end- 2.3. Stimulus conditions ing after sixteen trials, were interleavedinrandom order. Atotal of eight reference AM stimuli were used, four Foreach condition, the average LRBT of three repeats de- with acarrier frequencyof63Hzand four with 125 Hz. fined the individual’sLRBT,unless avalue deviated by We predicted that the infrasound detection might be eas- more than 4dBfrom anyother,inwhich case it wasex- ier at larger sound pressure, and modified therefore two cluded from the average. further parameters of the AM tone for which we expected The LRBT were then individually set for the final dis- the required biasing tone level(LRBT)tovary to produce crimination task, which wasa1-interval, yes-no proce- amatching perceivedloudness modulation of the biased dure with the question: “Did this stimulus contain infra- pure tone: the modulation depth of the AM tone (25% sound?” Foreach of the eight conditions, atotal of 50 tri- and 37.5%)and its sound pressure. All modulations were als were presented, where half consisted of the AM tone clearly audible. Since we expected asuppression in loud- and half of the corresponding biased tone (containing the ness due to biasing [7], the AM tones and pure tones to be infrasound at LRBT). This gave atotal of 400 trials, which

826 Marquardt, Jurado: Amplitude modulation confusion ACTA ACUSTICA UNITED WITH ACUSTICA Vol. 104 (2018) were presented in random order.The variation in loudness and modulation depth across the eight stimulus conditions 116 modulation depth made the task less monotonous for the subjects. Before the 37.5% actual test, the subjects had atraining session with feed- 114 25% back, using ashorter series of 48 trials. No feedback was

SPL) 112 givenduring the formal test. (dB T

RB 110 L

3. Results and discussion 108

106 Before looking at howwell listeners were able to distin- 125 Hz 63 Hz 125 Hz 63 Hz guish the AM stimuli from the stimuli that truly contained 50 phon 50 phon 40 phon 40 phon infrasound, let us first consider the infrasound levels that the listeners adjusted during the MLTprocedure so that the Figure 2. Average LRBT across subjects. perceivedmodulation of the biased tones matched that of the corresponding AM reference stimulus. The mean data across subjects for all 8stimuli condi- 90 Hits Correct rejections tions are shown in Figure 2. 80 70

According to the infrasound equal-loudness contours t

ec 60 proposed by Møller and Pedersen [5], LRBTsfell roughly orr in the range of 20–40 phon, and were therefore well above tc 50 cen 40 the 8-Hz perception threshold of most human listeners (ap- per 30 prox. 100 dB SPL[4]). Although the levels across the eight

mean 20 conditions stayed for each individual typically within a 10 range of 10 dB and resembled roughly the pattern of the 0 mean data, the individual curves were quite offset from S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 each other (For individual LRBT,see the table below.) As to expect, an increase in L is required to achieve Figure 3. Average discrimination performance across all eight RBT conditions for each of the 12 subjects. The length of each an increased perceptual modulation that matches the in- percent-correct bar is divided according to the contributions of crease of modulation depth from 25% to 37.5% in the ref- hits and correct rejections to the total of the correct responses. erence AM stimuli. Apart from this offset, the twocurves look rather similar across the remaining parameter varia- tions. Similarly expected, the louder 50-phon probe tones at amore basal and stiffer BM location. However, this is require generally more intense infrasound tones than the only the case with 40-phon carrier tones. softer 40-phon probe tones. We therefore conclude that perceptual modulation of an It has to be pointed out, however, that asimple linear LF tone during infrasound exposure cannot be simply ex- superposition model, as illustrated in Figure 1, does not plained by linear superposition of the two. We believe that quantitatively account for these results: A25% modula- large cochlear microphonic potentials, experimentally ob- tion depth in the half-waverectified output signal would served by Salt et al. [8] in guinea pigs, might cause an require aBM-biasing amplitude that is 25% of the probe “electrical” biasing of the inner hair cell neurotransmitter tone BM response amplitude. Similarly,a37.5% mod- release, leading to amodulation of the AN output. The ulation depth would require aBM-biasing amplitude of involved electrical phenomena are likely very non-linear, 37.5% of the probe tone BM response amplitude. This the- and only adetailed model of these rather complexpro- oretical 3.5-dB increase in LRBT is not observed for the 40- cesses might have the potential to quantitatively explain phon data. The required increase in BM biasing amplitude the present results, as well as previously published data on would also just linearly scale up with probe tone amplitude modulation of the AN activity by low-frequencybiasing if the perceivedmodulation could be simply explained by tones (e.g. [9, 10, 11]). linear superposition of the twotones. In other words, the Although the adjustment of biasing tone levels wasa roughly 7-dB increase in probe tone level(from 40 to 50 prerequisite to address the main question of this study, phon)should then also require aroughly 7-dB increase the data so farhavenot yet answered whether listeners in LRBT in order to maintain the same amplitude modu- can pick out the stimuli that contain the infrasound tone. lation of the half-waverectified output signal. However, Discrimination results per condition showed that the num- this is only the case for the 63-Hz condition with 37.5% ber of subjects scoring above 64% wasalways about half. modulation depth, butwas clearly lower in the other con- Note that only scores above 64% have achance of cor- ditions. This brings us to the last, rather puzzling obser- rectly guessing below5%(n=50). In other words, in each vation: We expected that the 125-Hz carrier tone requires condition about half of subjects performed no better than aslightly higher biasing tone levelthan the lower 63-Hz chance. Takenthe performance across all eight conditions carrier tone, because its excitation centroid being located (Figure 3),however,most subjects performed well above

827 ACTA ACUSTICA UNITED WITH ACUSTICA Marquardt, Jurado: Amplitude modulation confusion Vol. 104 (2018)

that none of the listeners picked the stimuli containing the 70 HitsCorrect rejections infrasound tone with great confidence. Comparing the eight stimulus conditions, none of them 60

t appeared to be particularly hard or easy: As Figure 4 ec 50 shows, the across-subject mean scores ranged between orr

tc 40 60% and 64.3%. However, the pattern across the condi- cen tions wasinconsistent amongst the listeners. In a3-way per 30 ANOVA ,neither modulation depth, carrier frequency, nor

mean 20 carrier levelturned out to be asignificant factor (p = 0.58, 0.97, 0.87 and F = 0.31, 0.0, 0.03, respectively). This 10 wasalso the case when only considering the six best per- 0 forming subjects. In other words, the easiest condition for Mod. depth (%)37.5 37.5 37.537.5 25 25 25 25 one subject wasone of the hardest conditions for another. Level(phon) 50 50 40 40 50 50 40 40 Frequency (Hz) 12563125 63 12563125 63 Avg. Also surprising, there wasnocorrelation of infrasound de- 2 LRBT 112114 109110 108 111 105107 110 tectability with infrasound SPLwhen pooling all data (R S1 N 8412 5781178 Hits = 0.00082, p >0.05). Individually,however,there were NCR 15 18 15 18 15 14 18 18 16

LRBT 111 107 107 106111 106 105103 107 cases of significant positive as well as negative correla- S2 N 13 18 17 19 18 16 17 20 17 Hits tions, possibly reflecting different listening strategies. NCR 17 17 19 15 12 20 14 17 16

LRBT 114 115 110 109108 108 105102 109

S3 NHits 13 16 16 16 17 15 19 21 17

NCR 20 18 20 20 23 19 20 18 20 4. Conclusion

LRBT 115 114 107 108109 109 105102 109 S4 N 18 13 14 18 17 17 22 15 17 Hits It has been shown under laboratory conditions that a NCR 18 18 15 14 18 21 18 14 17 LRBT 116 118 112 113113 115 110110 113 low-frequencytone that is 8-Hz amplitude-modulated is S5 N 19 23 20 22 19 22 21 23 21 Hits perceptually similar to astimulus that contains alow- NCR 21 19 18 17 20 21 21 19 20

LRBT 117 116 114 112113 114 112109 113 frequencytone and an 8-Hz infrasound tone. We specu- S6 N 14 914991212911 Hits late that other slowly amplitude-modulated low-frequency NCR 24 25 25 24 25 25 24 23 24

LRBT 119 116 116 111115 111 114111 114 stimuli, which actually do not contain spectral content be- S7 N 69913 12 9879 Hits low20Hz, might also sound as theywould contain in- NCR 13 16 17 81114101513

LRBT 109 108 105 99 108102 103103 105 frasound. This finding may help to explain cases of an- S8 N 12 16 13 12 10 12 17 15 13 Hits noyance attributed to infrasound, where measurements N 13 19 18 20 21 15 15 17 17 CR showaudible low-frequencycontent, butwith infrasound LRBT 118 121 117 110115 120 114108 115 S9 N 14 17 15 13 14 17 14 16 15 Hits content well belowsensation threshold. Nevertheless, N 14 15 12 81512171013 CR giventheir perceptual similarity,LFsound that is slowly LRBT 115 113 110 96 112112 109102 109

S10NHits 14 16 18 16 15 17 15 17 16 amplitude-modulated can still cause annoyance, similar to N 10 13 12 91513915 12 CR that attributed to true infrasound. LRBT 122122 120117 121 119 117114 119

S11NHits 18 20 17 19 18 18 19 20 19 N 17 18 20 22 17 20 20 19 19 CR Acknowledgements LRBT 120 119 117 114117 116 116113 117 S12N 12 12 15 15 17 12 11 12 13 Hits This study wassupported by the European Metrology Pro- N 17 16 14 13 15 13 14 14 15 CR gramme for Innovation and Research (EMPIR, Grant No. Figure 4. The table header also labels the percent correct bars, 15HLT03 Ears II). EMPIR is jointly funded by the EMPIR which showthe across-subject average score in each of the eight participating countries within EURAMET and the Euro- conditions. The length of each bar is divided proportional to the pean Union. We thank Jasdeep Benning for helping with number of hits (NHits)and correct rejections (NCR). the data collection.

References chance, since theyscored in the average better than 55 % (p <0.05 for n = 400). The best performance wasthat [1] C. S. Pedersen, H. Møller,K.Persson Waye: Adetailed of listener 5, with an average of 81% correct responses study of low-frequencynoise complaints. J. LowFreq. Noise Vib. 27 (2008)1–33. across the eight conditions. He is one of the authors, and performed at this levelsimilarly across all 8conditions. In [2] B. C. J. Moore: Contributions of vonBékésy to psychoa- coustics. Hear.Res. 293 (2012)51–57. his best condition, he achieved86% correct responses. But note that this does not necessarily mean that he detected [3] C. Jurado, B. C. J. Moore: Frequencyselectivity for frequencies below100 Hz: comparisons with mid- the infrasound in 86% of the trials, as the percentage also frequencies. J. Acoust. Soc. Am. 128 (2010)3585–3596. includes correct guesses with a50% chance. On the other [4] R. Kuehler,T.Fedtke, J. Hensel: Infrasonic and low- end, subject 1was amongst the worst performers, respond- frequencyinsert earphone hearing threshold. J. Acoust. ing with 48% correct responses purely at chance. He is the Soc. Am. 137 (2015)EL347–EL353. other author and had also long experience in listening ex- [5] H. Møller,C.S.Pedersen: Hearing at lowand infrasonic periments with infrasound. In summary,itisfair to say frequencies. Noise Health 6 (2004)37–57.

828 Marquardt, Jurado: Amplitude modulation confusion ACTA ACUSTICA UNITED WITH ACUSTICA Vol. 104 (2018)

[6] ISO 226:2003: Acoustics –Normal equal-loudness con- [9] Y. Cai, C. D. Geisler: Suppression in auditory-nervefibers tours. International Organization for Standardization, of cats using low-side suppressors. I: Temporal aspects. Geneva,Switzerland. Hear Res. 96 (1996)94–112. [7] E. Zwicker: Masking-period patterns produced by very- [10] A. N. Temchin, N. C. Rich, M. A. Ruggero: Low-frequency low-frequencymaskers and their possible relation to basi- suppression of auditory nerveresponses to characteristic lar-membrane displacement. J. Acoust. Soc. Am. 61 (1977) frequencytones. Hear.Res. 113 (1997)29–56. 1031–1040. [8] A. N. Salt, J. T. Lichtenhan, R. M. Gill, J. J. Hartsock: [11] H. Nam, J. J. Guinan: Low-frequencybias tone suppression Large endolymphatic potentials from low-frequencyand of auditory-nerveresponses to low-levelclicks and tones. infrasonic tones in the guinea pig. J. Acoust. Soc. Am. 133 Hear.Res. 341 (2016)66–78. (2013)1561–1571.

829