| [Download PDF]
|Year : 2005 | Volume
| Issue : 27 | Page : 1--10
Sound source identification with ANR earmuffs
SM Abel, JE Shelly Paik
Defence Research and Development, Toronto, Canada
S M Abel
Communications Group, Human Factors Research and Engineering Section, Defence Research & Development Canada - Toronto, P.O. Box 2000, 1133 Sheppard Ave. W. Toronto, Ontario, M3M 3B9
The effect of hearing protective earmuffs which incorporate active noise reduction (ANR) on sound source identification was studied. The purpose was determine whether ANR interfered with the encoding of cues normally used for directional hearing. Right/left, front/back and within quadrant confusions were assessed in quiet using a circular array of eight loudspeakers. Three stimuli, one-third octave bands centered at 0.5 kHz and 4 kHz and broadband noise, were presented. These enabled an assessment of the utilization of mainly interaural time-of-arrival and level differences, and binaural and spectral cues in combination, respectively. Two groups of normal hearing subjects aged 18-30 and 40-55 years, half male and half female, participated. Overall, age, gender, and ANR were not significant determinants of outcome. The probably of correctly discriminating among the eight speakers decreased significantly with the muffs worn, relative to unoccluded listening by 10%, 35% and 40% for the 0.5 kHz, 4 kHz and broadband stimuli, respectively. The pattern of errors indicated that the earmuffs interfered with the encoding of both binaural (interaural level differences) and spectral cues. With ANR small additional right/left confusions were observed for the low-frequency stimulus (time-of arrival cue) for speakers close to the midline axis. The results provide further evidence that earmuffs should not be used in situations where the perception of the direction of hazard is a concern. ANR technology does not appear to increase the handicap.
|How to cite this article:|
Abel S M, Shelly Paik J E. Sound source identification with ANR earmuffs.Noise Health 2005;7:1-10
|How to cite this URL:|
Abel S M, Shelly Paik J E. Sound source identification with ANR earmuffs. Noise Health [serial online] 2005 [cited 2021 Jan 21 ];7:1-10
Available from: https://www.noiseandhealth.org/text.asp?2005/7/27/1/31637
The purpose of this study was to investigate the effect of wearing a hearing protection ear muff incorporating active noise reduction (ANR) on sound source identification. Previous research has shown that conventional sound-attenuating earplugs and muffs, worn in either a quiet or noisy background, results in decrements in the ability to distinguish right from left and front from back, relative to unoccluded listening (Atherley and Noble, 1970; Russell, 1976; Noble et al., 1990). Level-dependent muffs with limited amplification have been shown to be more detrimental to right/left discrimination than conventional muffs. The two types of muff also differ with respect to error patterns in front/back discrimination (Abel and Hay, 1996). Right/left errors signify disruption of mainly interaural time-of-arrival (below 1.5 kHz) and level (above 1.5 kHz) difference cues normally used by the brain for free field sound source identification. Front/back errors signify disruption of spectral cues provided by the outer ear (Musicant and Butler, 1984; Blauert, 1997).
To date, there have been no investigations of the effect on sound source identification of hearing protection muffs which incorporate active noise reduction (ANR) technology. ANR hearing protectors are of particular interest in military applications because they have the capability of enhancing low-frequency sound attenuation to a greater extent than possible with conventional passive ear muffs. A microphone housed within the ear cup samples the internal noise and adds it out of phase, thereby diminishing the overall level reaching the eardrum. ANR hearing protection devices increase attenuation for frequencies below 1 kHz, by amounts of up to 20 dB, depending on the device selected (McKinley et al., 1996). This has the additional benefit of improving speech understanding because of the reduction of the upward spread of low-frequency masking noise to the higher speech frequencies (Abel et al., 1990).
The goal of the present study was to assess the effect on sound source identification in adults with no more than a mild hearing loss of an earmuff which incorporates ANR technology. A unique feature of the device selected was the venting of the ear shells. Venting may promote barometric equalization in the ear canal, thereby relieving the feeling of pressure (Lybarger, 1985). The specific aims were to provide new information on the ability to use auditory cues, namely, interaural differences in time-of-arrival and level of sound, as well as binaural and spectral cues in combination to localize hazard in situations that require the wearing of ANR hearing protection devices. The effects of aging and gender were also assessed. Previous studies have demonstrated that even with normal hearing, subjects' accuracy in localizing sound sources decreases as they age, beginning in the fourth decade of life (Abel et al., 2000). Women may achieve less attenuation than men with both muffs and plugs, owing to the smaller size of the head and ear canal (Abel et al., 1988; Abel et al., 2002).
Methods and Materials
The study protocol was approved by the Defence Research and Development Canada Human Research Ethics Committee. Two groups of 12 normal hearing adults, aged 18-30 years (Y) and 40-55 years (O) were tested to allow an assessment of the effect of age. Half the subjects in each group were males (M) and half were females (F) to enable the study of gender effects. Normal hearing was defined as a hearing threshold no greater than 25 dB HL from 0.5-4 kHz (Yantis, 1985), and an interaural difference in hearing no greater than 10 dB averaged across the frequencies tested. It has been shown that hearing loss does not affect accuracy in unoccluded sound localization, as long as the subject can perceive the stimulus (Noble et al., 1994). The constraint on the difference in hearing between the two ears minimized a possible bias in right/left discrimination.
Subjects were tested individually while seated in the centre of a circular array of eight loudspeakers situated within a double-walled semi-reverberant sound proof booth (Giguere and Abel, 1990). Two loudspeakers were positioned in each of the four spatial quadrants, one close to the midline axis and the second close to the interaural axis of the head at the following azimuth angles: 15°, 75°, 105°, 165°, 195° (-165°), 255° (-105°), 285° (-75°) and 345° (-15°). They were chosen to be right/left and front/back symmetric to enable an assessment of mirror image, as well as between and within-quadrant confusions. The speakers were placed at a distance of 1 m from the subject's centre head position at ear height. Accuracy in sound source identification and the time to respond were measured with the ears unoccluded and protected with a representative ANR earmuff, with and without ANR operational. In the latter condition, the device behaved like a conventional earmuff, that is, the sound attenuation at a given frequency did not change with changes in sound level.
Three stimuli were used to measure accuracy within each of the three ear conditions: two one third octave noise bands centered at 0.5 kHz and 4 kHz, and broadband noise. These allowed an independent evaluation of the utilization of mainly interaural differences in time-of-arrival and level and binaural and spectral cues in combination, respectively (Blauert, 1997; Abel et al., 2000). Noise bands were used in preference to pure tones to avoid local minima and maxima in sound level that might result from the small degree of room reverberation. The stimuli were 300 ms in duration (including a linear rise/decay time of 50 ms) and were presented at a comfortable listening level of 75 dB SPL (sound pressure level) that was clearly audible without and with the protectors worn. Subjects were not given feedback about the correctness of their judgments. Response time, an index of task difficulty (Vickers, 1980) was measured covertly, so as not to affect accuracy (Link,1978; Luce, 1986).
The subjects were recruited for the study by posting notices at Defence Research and Development Canada - Toronto (DRDC Toronto). Each provided written informed consent. It was determined that none of the subjects had previously participated in a sound localization experiment. All underwent a hearing screening test to ensure that the hearing criteria were met. In all but five subjects, hearing thresholds in each ear were no greater than 15 dB HL at 0.5, 1, 2 and 4 kHz. In the five outliers, thresholds ranged from 16-26 dB HL.
The test facility was a double-walled semi-reverberant sound proof booth (IAC Series 1200) with inner dimensions of 3.5 (L) X 2.7 (W) X 2.3 (H) metres that met the standard for hearing protector testing (ANSI 1997). Reverberation times were 0.6 s at 0.125 kHz and 0.25 kHz, 0.4 s from 0.5 kHz to 4 kHz, and 0.3 s at 6.3 kHz and 8 kHz, indicating that the room was fairly absorbent. Ambient noise in the booth was less than the maximum allowed for audiometric testing (ANSI, 1996). The stimuli to be localized were generated by a noise generator (Bruel & Kjaer Type 1405) and one-third octave band pass filter (Bruel & Kjaer Type 1617). Stimulus envelope shape and duration and trial by loudspeaker selection were controlled by a Coulbourn Instruments modular system. Level was specified using a programmable attenuator (Coulbourn S85-08) and a set of integrated stereo amplifiers (Realistic SA-150). The stimuli were presented by a set of eight loudspeakers (Radio Shack Minimus 3.5) closely balanced with respect to output levels. Subjects signified their spatial judgments by means of a specially designed laptop response box with a set of eight raised micro-switches in the same configuration as the speaker array, both in number of elements and azimuth angles. Response times could be measured with a 1-msec accuracy. Detailed descriptions of the booth, instrumentation and calibration methods for the sound source identification test have been published previously (Giguere and Abel, 1990).
Each subject participated in one test session, lasting approximately 75 minutes. At the start of the session, the subject was instructed to sit squarely and to try to minimize head movement by using a small adjustable headrest affixed to the back of the chair. Head movement may help to resolve front/back confusions (Wightman and Kistler, 1999). A set of two practice trials/loudspeaker with feedback (i.e., 16 trials) using the broadband stimulus was given at the start of the experiment to provide the subject with a spatial sense of the loudspeaker array relative to the response buttons, and to ensure that the instructions had been understood.
One block of 120 forced-choice loudspeaker identification trials was given for each of the nine experimental conditions (3 ear conditions x 3 stimuli). The unoccluded condition was presented first followed by the two protected conditions, ANR Off and ANR On. Orders of the two protected conditions, and three stimuli within ear condition were counterbalanced across the six subjects in each of the four groups. A trial block comprised 15 random presentations of the stimulus through each of the eight loudspeakers.
Each trial began with a ˝-s warning light on the response box, followed by a ˝-s delay and then the presentation of the 300-ms stimulus. The stimulus duration incorporated a 50-ms rise/decay time to minimize the effect of onset. The warning light was the subject's cue to focus on a straight-ahead visual target mounted on the wall of the booth. This ensured the alignment of the speaker array and the coordinate system of the head. The rate of presentation of trials was approximately one every seven seconds. Following each stimulus presentation, the subject was required to push the micro-switch on the laptop response box corresponding to the loudspeaker that had emitted the stimulus.
Guessing was encouraged and no feedback was given about the correctness of the judgments. Subjects were advised to use both hands for responding, the right hand for response buttons on the right side and the left hand for response buttons on the left side to minimize an increase in response time from crossing the hand to the contralateral side.
Both percent correct sound source identification judgments and the median response time were calculated for each block of trials. The median response time was chosen in preference to the mean because of previously demonstrated skewness in response time data (Abel et al., 1990). The effects on sound source identification of group, ear condition (unoccluded, ANR On and Off), and stimulus frequency (0.5 kHz, 4 kHz, broadband) were assessed using a repeated measures analysis of variance (ANOVA) (Daniel, 1983). All statistical analyses were performed using SPSS computer software.
[Table 1] shows the mean percent correct, P(C), sound source identification judgments for each group (YM-young males, YF-young females, OM-older males, OF-older females) for combinations of ear condition and stimulus frequency. [Figure 1] shows these same data averaged across groups. The ANOVA indicated that there were significant effects of ear condition [F(2,40)=326.6; pQuadrant P(C)
[Table 2] shows the P(C) associated with identification of the spatial quadrant from which the sound emanated, regardless of loudspeaker, averaged across groups. The pattern of outcomes for each quadrant was similar to that for overall P(C). The ANOVA applied to these data indicated that there were significant effects of ear condition [F(2,40)=245.7; p Azimuthal P(C)
[Figure 2] shows the effect of ear condition on the P(C) achieved for each of the eight azimuth angles, for the three stimuli respectively, averaged across the four groups. The ANOVA indicated that there were significant effects of ear condition [F(2,40)=326.6; pError Analysis
[Table 3] shows the percentages associated with various error types for combinations of ear condition and stimulus frequency. The data have been collapsed across group and azimuth angle. The error types examined included RE, a front/back mirror image reversal error; QE, a within-quadrant error; SE, a within-side error other than an RE or QE; and ME, an error on the opposite side of the midline (Abel et al., 2000). Regardless of ear condition or stimulus frequency, the most prevalent error was a front/back mirror image reversal error. An ANOVA applied to the REs indicated that there were significant effects of ear condition [F(2,40)=47.1; pMedian Response Time
[Table 4] shows the median response time for sound source identification judgments for the four groups. An ANOVA applied to these data indicated that there were significant effects of ear condition [F(2,40)=5.6; p<0.007] and stimulus frequency [F(2,40)=9.0; p<0.001]. Averaged across the four groups and three stimuli, mean median response times ranged from 712 ms for ANR On to 805 ms in the unoccluded condition. Averaged across the four groups and three ear conditions, mean median response times ranged from 667 ms for the broadband noise to 803 ms for the 0.5 kHz stimulus. Post hoc pair wise comparisons indicated that although the overall trends were significant there were no significant differences among pairs of outcomes for the three ear conditions or three stimuli.
This study was carried out to determine whether the operation of active noise reduction in a representative sound attenuating earmuff with ANR capability would negatively affect sound source identification ability in normal-hearing adults, relative to conventional passive attenuation and unoccluded listening. Previous studies had documented the detrimental effects of wearing conventional hearing protective earplugs and earmuffs, as well as additional decrements in accuracy with limited amplification (Atherley and Noble, 1970; Abel and Hay, 1996). The results of the present investigation showed clearly that there was no effect of ANR on overall P(C). Both modes of operation (ANR Off and ANR On) resulted in significantly less accurate discrimination among loudspeakers than observed with the ears unoccluded. Other factors assessed were the age and gender of subjects. The results showed that group (age by gender) was not a significant determinant of outcome. This was in contrast to previous research which had shown a decrease of approximately 12-15% in the correct discrimination of surround sound sources between the ages of 10 and 80 years (Abel et al., 2000). The latter effect was largely attributable to a progressive deficit in the processing of spectral information which accelerated after the sixth decade of life. The average ages of subjects in the young and older groups in the present study were 24 years and 48 years.
The analyses undertaken also generated new quantitative information for the earmuff conditions about the processing of cues normally used by the brain for sound localization when earmuffs are worn. To correctly identify the location of the sound source in the eight-speaker array, subjects were required to answer three questions, each with two possible answers (Abel et al., 2000). The answers to two of these questions, right versus left and front versus back, determined the spatial quadrant. Subjects would then have to decide whether the sound source was close the midline or interaural axis within the quadrant identified. Using this scheme, the probability of a correct response is by chance 0.125 (0.5 x 0.5 x 0.5). Our previous research has shown that regardless of the stimulus, with the ears unoccluded subjects have no difficulty in determining side of space and azimuth within quadrant for the array used in the present experiment. Thus, they can obtain at least 50% correct. The main difficulty is front/back discrimination which depends on the availability of spectral cues provided by the pinnae of the ears (Giguere and Abel, 1993). As the angle between the sound source and the ear changes, so too will the spectral composition of the sound (Russell, 1976). Spectral cues are particularly salient for frequencies beyond 3 kHz (Musicant and Butler, 1984).
In the unoccluded condition, overall accuracy, that is the percent correct sound source identification judgments averaged across the eight azimuth angles tested, was 51% with the 0.5 kHz stimulus. With this stimulus, subjects have access mainly to the interaural time-of arrival cue. This allows discrimination of sound sources on the right and left sides of space and the discrimination of sound sources within a spatial quadrant but not front versus back. By comparison, accuracy increased significantly to 59% for the 4 kHz stimulus which is localized using mainly the interaural level difference and to some degree spectral information, and to 91%with the broadband noise which is localized using binaural and spectral cues in combination. These results replicated previous findings (Abel et al., 2000). When the ear muff was donned, accuracy in identifying the speaker that had emitted the 0.5 kHz stimulus decreased by 7%, signifying that either right/left or within quadrant discrimination was affected. An additional decrement of 6% was observed when ANR was operational. With the 4 kHz and broadband stimuli, the decrements in accuracy with the muff were 35% and 40%, respectively, signifying serious disruption of both the interaural level cue and binaural and spectral cues in combination. There was no additional effect of active noise reduction. The time taken to make a decision was not affected by the wearing of the device, suggesting that the difficulty of the task had not changed (Vickers, 1980).
The pattern of outcomes in each spatial quadrant mimicked the pattern observed above for overall P(C). There were no differences in quadrant P(C) for the right and left sides of space, front or back, although P(C) was consistently relatively greater on the left than the right. Left-sided superiority has been demonstrated previously (Burke et al., 1994; Abel et al., 2000). By comparison, there were many conditions in which the P(C) associated with the frontal quadrant significantly exceeded that for the back quadrant on both the right and left sides of space, and no conditions in which sound source identification was significantly better in back than in front. Similar effects have been noted previously for unoccluded ears and ears protected with conventional muffs (Abel et al., 2000; Russell and Noble, 1976). In contrast, rearward shifts in the auditory percept have been noted when muffs with limited amplification were worn (Abel and Hay, 1996).
Investigation of the P(C) associated with each of the eight azimuth angles revealed that for the 0.5-kHz stimulus which was localized primarily using the interaural time-of-arrival cue, the effect of the muff was limited to the azimuthal positions in front, on either side of the midline axis (i.e., 15 deg and 345 deg). These were the only conditions in which accuracy was somewhat worse with ANR operational. The error analysis showed that only the likelihood of an opposite-side error increased demonstrably, from 2% in the unoccluded condition to 4% in the ANR Off condition to 8% in the ANR On condition. For the 4-kHz stimulus, localized mainly using the interaural level difference cue, the probably of a correct response was approximately 20% regardless of the azimuth angle. The error analysis indicated that the incidence of within-quadrant, within-side, and opposite-side errors but not front/back mirror image reversal errors had all increased substantially, by 12% relative to unoccluded listening. For the broadband noise stimulus, localized using binaural and spectral cues in combination, the muff interfered with azimuths in back to a greater degree than those in front. Accuracy in identifying sound sources located on either side of the midline in back decreased by about 70% relative to unoccluded listening. The error analysis showed that only front/back mirror image reversal errors increased appreciably, by 36%.
Taken together, the results of the study argue strongly against the use of earmuffs in situations where the perception of the direction of hazard is a concern. In general, implementation of active noise reduction neither facilitated nor further degraded sound source identification ability relative to listening with a conventional muff. However, small additional right/left confusions were observed with ANR for the low-frequency stimulus (time-of-arrival difference cue) in the case of speakers located close to the midline. The muffs mainly decreased accuracy in localizing the 4-kHz (level difference cue) and broadband (binaural and spectral cues in combination) stimuli. This led to an increase in within-quadrant and right/left and front/back errors in sound source identification, respectively.
This research was funded by the Military Operational Medicine Thrust, Defence Research and Development Canada-Toronto.
|1||Abel, S.M., Alberti P.W., and Rokas D. (1988). Gender differences in real-world hearing protector attenuation. J. Otolaryngol. 17(2): 86-92.|
|2||Abel, S.M., Giguere, C, Consoli, A., and Papsin, B.C. (2000). The effect of aging on horizontal plane sound localization. J. Acoust. Soc. Am. 108(2): 743-752.|
|3||Abel, S.M., and Hay, V.H. (1996). The interaction of aging, hearing loss and hearing protection. Scand. Audiol. 25: 3-12.|
|4||Abel, S.M., Krever, E.M., and Alberti, P.W. (1990). Auditory detection, discrimination and speech processing in ageing, noise-sensitive and hearing-impaired listeners. Scand. Audiol. 19(43): 43-54.|
|5||Abel, S.M., Rajan, D.K. and Giguere, C. (1990). Stimulus parameters in detection and reaction time. Scand. Audiol. 19: 229-235.|
|6||Abel, S.M., Sass-Kortsak, A., and Kielar, A. (2002). The effect on earmuff attenuation of other safety gear worn in combination. Noise & Health 5: 1-13.|
|7||ANSI (1996). ANSI S3.6-1996. Specifications for audiometers. American National Standards Institute, New York.|
|8||ANSI (1997). S12.6-1997. Methods for measuring the real-ear attenuation of hearing protectors. American National Standards Institute, New York.|
|9||Atherley, G.R.C., and Noble, W.G. (1970). Effect of eardefenders (earmuffs) on the localization of sound. Br. J. Ind. Med. 27: 260-265.|
|10||Blauert, J. (1997). Spatial Hearing. MIT Press, Cambridge.|
|11||Burke, K.A., Letsos, A., and Butler, R.A. (1994). Asymmetric performance in binaural localization of sound in space. Neuropsychol. 32: 1409-1417.|
|12||Daniel, W.W. (1983). Biostatistics: A Foundation for Analysis in the Health Sciences. Wiley, New York.|
|13||Giguere, C., and Abel, S.M. (1990). A multi-purpose facility for research on hearing protection. Appl. Acoust. 31: 295-311.|
|14||Giguere, C., and Abel, S.M. (1993). Sound localization: effects of room reverberation, speaker array, stimulus frequency and stimulus rise/decay. J. Acoust. Soc. Am. 94(2). Pt. 1: 769-776.|
|15||Link, S.W. (1978). The relative judgment theory of response time deadline experiments. In Cognitive theory III. Castellan, N.J. Jr., & Restle, F., eds. Erlbaum, Hillsdale, pp 117-138.|
|16||Luce, R.D. (1986). Response times: their role in inferring elementary mental organization. Oxford University Press, New York.|
|17||Lybarger, S.F. (1985). Earmolds. In Handbook of Clinical Audiology. Katz, J., ed. Williams and Wilkins, Baltimore, pp 885-910.|
|18||McKinley, R.L, Steuver, J.W., and Nixon, C.W. (1996). Estimated reductions in noise-induced hearing loss by application of ANR headsets. In Scientific Basis of NoiseInduced Hearing Loss. Axelsson, A., Borchgrevink, H., Hamernik, R.P., Hellstrom, P.-A., Henderson, D., and Salvi, R.J., eds. Thieme, New York, pp 347-360.|
|19||Musicant, A.D., and Butler, R.A. (1984). The influence of pinnae-based spectral cues on sound localization. J. Acoust. Soc. Am. 75:1195-1200.|
|20||Noble, W., Byrne, D., and Lepage, B. (1994). Effects on sound localization of configuration and type of hearing impairment. J. Acoust. Soc. Am. 95: 992-1005.|
|21||Noble, W., Murray, N. and Waugh, R. (1990). The effect of various hearing protectors on sound localization in the horizontal and vertical planes. Am. Ind. Hyg. Assoc. J. 51(7): 370-377.|
|22||Russell, G. (1976). Effects of earmuffs and earplugs on azimuthal changes in spectral patterns: implications for theories of sound localization. J. Aud. Res. 16:193-207.|
|23||Russell, G., and Noble, W.G. (1976). Localization response certainty in normal and disrupted listening conditions: toward a new theory of localization. J. Aud. Res. 16:143150.|
|24||Vickers, D. (1980). Discrimination. In Reaction Times. Welford, A.T., ed. Academic, New York, pp 25-72.|
|25||Wightman, F.L., and Kistler, D.J. (1999). Resolution of front-back ambiguity in spatial hearing by listener and source movement. J. Acoust. Soc. Am. 105: 2841-2853.|
|26||Yantis, R. A. (1985). Pure tone air-conduction testing. In Katz, J. (ed.) Handbook of Clinical Audiology, 3 r d ed. Katz, J. ed. Williams & Wilkins, Baltimore, pp. 153-169.|