Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Speech perception wikipedia , lookup
Telecommunications relay service wikipedia , lookup
Sound from ultrasound wikipedia , lookup
Auditory system wikipedia , lookup
Hearing loss wikipedia , lookup
Hearing aid wikipedia , lookup
Sensorineural hearing loss wikipedia , lookup
Sound localization wikipedia , lookup
Noise-induced hearing loss wikipedia , lookup
Audiology and hearing health professionals in developed and developing countries wikipedia , lookup
The effects of single and double hearing protection on the localization and segregation of spatially-separated speech signals (L) Douglas S. Brungart Air Force Research Laboratory, WPAFB, Ohio Alexander J. Kordik Sytronics, Dayton, Ohio Brian D. Simpson Air Force Research Laboratory, WPAFB, Ohio 共Received 25 December 2003; revised 21 May 2004; accepted 9 July 2004兲 Recent results have shown that auditory localization in the horizontal plane is dramatically worse for listeners wearing double hearing protection 共earplugs and earmuffs兲 than it is for listeners wearing single hearing protection 共earplugs or earmuffs alone兲. This suggests that double hearing protection might also impair the spatial unmasking that normally occurs when two simultaneous talkers are spatially separated in azimuth 共the so-called ‘‘cocktail party’’ effect兲. In this experiment, normal hearing listeners wearing no hearing protection, single hearing protection 共earmuffs兲, or double hearing protection were asked to perform a speech intelligibility task that required them to segregate two simultaneous talkers who were either presented from the same loudspeaker or spatially separated by 90° in azimuth. The listeners were also asked to determine the location of the target talker in each trial. The results show that the listeners were unable to reliably determine the location of the target talker when they wore double hearing protection, but that they were still able to benefit from the spatial separation of the competing talkers. This suggests that the use of double hearing protection causes spatially separated sound sources to be heard at locations that are inaccurate but still distinct enough to enhance the segregation of speech. 关DOI: 10.1121/1.1786812兴 PACS numbers: 43.50.Hg, 43.66.Vt, 43.66.Qp 关AR兴 I. INTRODUCTION Hearing protection devices play an essential role in preventing long-term occupational hearing loss, but they can also impair the localization cues that listeners use to process and analyze surrounding sound sources. This is especially true in environments with noise levels that exceed 100 dBA, where double hearing protection 共a combination of earplugs and earmuffs兲 should be worn to avoid the possibility of permanent hearing loss 共NIOSH, 1998兲. Recent results from our laboratory have shown that listeners wearing double hearing protection are much worse at sound localization than those wearing earplugs alone or earmuffs alone. Earplugs and earmuffs have long been known to impair the spectral cues used for the localization of brief sounds in the horizontal plane 共Abel and Hay, 1996; Abel and Armstrong, 1993; Vause and Grantham, 1999; Bolia, D’Angelo, Mishler, and Morris, 2001兲, but most studies have shown that listeners wearing single hearing protection can adequately localize sound sources in azimuth when they are on long enough to facilitate the use of exploratory head movements 共Noble, Murray, and Waugh, 1990兲. However, two recent studies in our laboratory have shown that listeners wearing double hearing protection are reduced to near chance localization performance in the horizontal plane even with continous sound sources that allow them to make unrestricted head movements 共Brungart, Kordik, Simpson, and McKinley, 2003; Simpson, Bolia, McKinley, and Brungart, 2002兲. J. Acoust. Soc. Am. 116 (4), Pt. 1, October 2004 Pages: 1897–1900 One possible explanation for the dramatic decrease in localization performance that occurs when double hearing protection is worn is that the interaural localization cues that are usually present in audio signals that enter the ears through the ear canals are being contaminated by bone- and tissue-conducted sounds that enter the listener’s ears directly through the surface of the head. This kind of bone conduction is already known to place a limit on the amount of attenuation that can be achieved with conventional earplug and earmuff hearing protection devices 共Zwislocki, 1957; Berger, Kieper, and Gauger, 2003兲. Once these devices attenuate the direct sound entering the ear canal below the level of the sound reaching the cochlea through bone or tissue conduction, additional attenuation no longer has any effect on the perceived level of the sound or on the long-term damage it can cause to the listener’s hearing. Researchers who have estimated the bone-conduction limit by measuring sound detection thresholds with increasingly effective earplug and earmuff combinations have found that the boneconduction limit on conventional hearing protection ranges from roughly 39 dB of attenuation at 2 kHz to roughly 55 dB of attenuation at 500 Hz 共Berger, 1983; Berger et al., 2003兲. Although bone-conduction pathways have traditionally been examined in the context of the limitations they impose on the effective attenuation of hearing protection devices, it is likely that these flanking pathways also influence how well listeners are able to localize sound sources while wearing hearing protection. Sound localization in the horizontal plane 0001-4966/2004/116(4)/1897/4/$20.00 1897 depends primarily on direction-dependent changes in interaural time and intensity differences caused by the diffraction of air-conducted soundwaves around the listener’s head and torso. These interaural difference cues would be badly corrupted in a bone-conducted sound, which would propagate directly through the head at a much higher rate than a normal air conducted signal. Consequently, it may not be surprising that a listener wearing double hearing protection that reaches the bone-conduction attenuation limit at frequencies at or above 1 kHz would perform much worse in a sound localization task than a listener wearing single hearing protection that does not approach the bone-conduction limit at any frequency. Indeed, similar reductions in localization ability have been reported for listeners with severe conductive hearing losses who experience the same kind of increase in the level of bone-conducted sound relative to the level of airconducted sound as listeners wearing double hearing protection 共Zurek, 1986; Noble, Byrne, and LePage, 1994兲. Although a reduction in sound localization ability is clearly one important consequence of the use of double hearing protection, it may not be the only consequence. The degradation in sound localization that occurs when double hearing protection is worn may also impact the processing of complex auditory scenes with multiple simultaneous sound sources. In normal listening environments, the ability to segregate and process multiple simultaneous sound sources is greatly enhanced when those sound sources are spatially separated. The classic example of this is the dramatic increase in speech intelligibility that occurs when a target speech signal is spatially separated from a masking speech signal, often referred to as the ‘‘cocktail party’’ effect 共Cherry, 1953兲. To the extent that double hearing protection disrupts the interaural localization cues that listeners use to segregate spatially separated sounds, one might expect the benefits of spatial separation inherent in the cocktail party effect to be greatly reduced in listeners wearing double hearing protection. Indeed, Zurek 共1986兲 has speculated that a similar reduction in spatial unmasking should occur in listeners with severe conductive hearing losses. If this kind of reduction in complex sound processing ability does occur in listeners wearing double hearing protection, then steps should be taken to alert the users of double hearing protection to this problem and to eliminate situations where a lack of spatial segregation ability could pose a safety hazard in high noise environments. The purpose of this study was to examine the effect that double hearing protection has on the segregation of multiple simultaneous speech sources. II. METHODS A. Subjects Six volunteer subjects, five males and one female, were paid to participate in the experiment. All had normal hearing 共⬍15 dB HL from 500 Hz to 8 kHz兲, and their ages ranged from 21 to 24 years. B. Hearing protection devices Three hearing protection 共HP兲 conditions were tested in the experiment: No HP, where no hearing protection device 1898 J. Acoust. Soc. Am., Vol. 116, No. 4, Pt. 1, October 2004 was used; Single HP, where the subjects wore earmuffs 共Tasco Soundshield兲; and Double HP, where the subjects wore fully inserted foam earplugs 共E•A•R Classic ®兲 under earmuffs 共TASCO Soundshield兲. In conditions where HP devices were used, they were fitted under the supervision of an experimenter. Prior to this study, the real ear attenuation at threshold method was used to determine the attenuation characteristics of the Single HP and Double HP hearing protection devices for each of the six participants in this experiment. These attenuation results, which are reported in detail elsewhere 共Brungart et al., 2003兲, indicate that the combination of earplugs and earmuffs used in the Double HP condition of this experiment reach the bone-conduction limit of attenuation for frequencies at or above 1 kHz, while the earmuffs used in the Single HP condition produce less attenuation than the bone conduction limit at all frequencies. C. Apparatus The experiment was conducted in the Air Force Research Laboratory’s Auditory Localization Facility 共ALF兲, a large anechoic chamber housing an aluminum-frame geodesic sphere with small loudspeakers mounted at each of its 272 vertices 共Brungart et al., 2003兲. These 272 loudspeakers were not capable of producing sufficiently intense auditory stimuli for the Double HP condition of this experiment. Consequently, the ALF was modified by mounting two large powered loudspeakers with twin 8-in. woofers 共Barbetta Sona兲 at ⫾45° azimuth in the horizontal plane 1.4 m away from the subject, who was positioned on a platform in the center of the sphere. These powered loudspeakers were used to generate all the stimuli used in the experiment. In order to allow the subjects to use a computer mouse to make their responses, a 14 in. color cathode-ray tube was also mounted outside the sphere between the two loudspeaker locations directly in front of the subject. D. Threshold measurement At the start of each block of trials, the subjects were fitted with the appropriate level of hearing protection and asked to stand in the middle of the sphere facing the midpoint between the two loudspeakers. Then they participated in a two-down, one-up adaptive threshold procedure 共Levitt, 1971兲 to estimate their hearing threshold level with that hearing protection device. In each trial of this two-interval forced-choice procedure, light emitting diodes located over the left and right buttons of a two-button response box were used to visually cue two consecutive 250-ms observation intervals, separated by a pause of 100 ms. A 250-ms burst of pink noise1 was presented in one of these two observation intervals, and the subject’s task was to determine which interval contained the noise burst. Two consecutive correct responses resulted in a 3 dB decrease in the noise level, and each incorrect response resulted in a 3 dB increase in the noise level. This procedure was repeated until ten reversals occurred, and the last five reversals were averaged to estimate the detection threshold for that particular subject with that hearing protection device. The mean detection thresholds measured with this procedure were roughly ⫺3 dB Brungart et al.: Letters to the Editor FIG. 1. The left panel shows performance in the left-right discrimination task, where the subjects had to identify which speaker the target phrase originated from. The right panel shows the percentage of correct color and number identifications in each condition of the experiment. The error bars represent the 95% confidence intervals calculated from all the raw data at each data point. sound pressure level 共SPL兲 in the No HP condition, 22 dB SPL in the Single HP condition, and 37 dB SPL in the Double HP condition.2 E. Procedure Once the threshold measurement was complete, data collection started in the speech intelligibility test. This test was based on the coordinate response measure 共CRM兲, a call-sign-based intelligibility test consisting of target and masking phrases of the form ‘‘Ready 共call sign兲 go to 共color兲 共number兲 now.’’ The phrases themselves were drawn from the publicly-available CRM corpus 共Bolia, Nelson, Ericson, and Simpson, 2000兲, which contains CRM phrases for all combinations of eight call signs 共‘‘Arrow,’’ ‘‘Baron,’’ ‘‘Charlie,’’ ‘‘Eagle,’’ ‘‘Hopper,’’ ‘‘Laker,’’ ‘‘Ringo,’’ ‘‘Tiger’’兲, four colors 共‘‘blue,’’ ‘‘green,’’ ‘‘red,’’ ‘‘white’’兲, and eight numbers 共1– 8兲 spoken by four male and four female talkers. In this experiment, the stimulus always consisted of two simultaneous CRM phrases, a target phrase randomly selected from all the phrases containing the call sign ‘‘Baron,’’ and a masking phrase randomly selected from all the phrases spoken by a different same-sex talker with a different call sign, color, and number than the target phrase. These phrases were adjusted to set their rms power levels 30 dB higher than the hearing threshold measured at the start of the block, and presented either from the same randomly selected loudspeaker 共‘‘nonspatial trials’’兲 or from different randomly selected loudspeakers 共‘‘spatial trials’’兲. In each case, the subject’s task was to first identify the color and number combination contained in the phrase containing ‘‘Baron,’’ and then to determine whether the target phrase came from the left or right loudspeaker. The subjects were permitted to move their heads during this procedure, but they were not specifically encouraged to do so. Each block consisted of 15 spatial and 15 nonspatial trials, and each subject participated in four blocks of trials with each of the three hearing protection levels, with the order of the blocks randomized across the different subjects. III. RESULTS The results of the experiment are shown in Fig. 1. The left panel of the figure shows performance in the left-right J. Acoust. Soc. Am., Vol. 116, No. 4, Pt. 1, October 2004 discrimination task of the experiment. These results show that left-right discrimination was near 100% correct in both the No HP and Single HP conditions, but that it fell to roughly 60% correct in the Double HP condition.3 A twofactor within-subjects analysis of variance 共ANOVA兲 on the arcsine-transformed data from each individual subject confirmed that this drop in performance was significant (F (2,10) ⫽27.95, p⬍0.0001). This finding is consistent with our earlier experiment that showed that listeners wearing double hearing protection are unable to reliably localize sound sources in the horizontal plane 共Brungart et al., 2003兲. Note that spatial separation of the talkers had no effect on leftright discrimination in any of the hearing protection conditions. Thus, it appears that the listeners were unable to reliably determine the location of the target talker in the Double HP condition even when both speech signals originated from the same loudspeaker and there was no need to use the target call sign to determine the location of the target phrase. The right panel of Fig. 1 shows performance in the color-number identification task of the experiment. In this task, spatial separation significantly improved performance in every hearing protection condition tested 共two-factor within subjects ANOVA on the individual arcsinetransformed scores兲 (F (1,5) ⫽211.63, p⬍0.0001). As in the left-right discrimination task, there was a significant decrease in performance in the Double HP condition (F (2,10) ⫽23.19, p⬍0.0002). There was also a significant 共15 percentage point兲 decrease in the magnitude of the spatial advantage in the Double HP condition (F (2,10) ⫽5.60, p⬍0.05). Nevertheless, despite the poor location discrimination data in the Double HP condition, the intelligibility advantages of spatial separation in that condition were substantial 共69% percent correct identifications spatial versus 38% non-spatial兲. On the surface, it seems somewhat odd that listeners could obtain such a large intelligibility advantage from spatial separation in a condition where they could not reliably distinguish between two spatial locations that were separated by 90° in azimuth. Indeed, the 60% correct discrimination scores achieved in the left-right discrimination task, while technically better than chance, were lower than what is generally considered to be the discrimination threshold for the minimum audible angular change in the location of a sound source, or MAA 共Mills, 1958兲. The most likely explanation for the spatial unmasking in the Double HP condition is that, although listeners in that condition could not accurately determine the actual locations of the talkers, they could tell that the talkers were originating from different locations, and this helped them to segregate the two speech signals into different streams and improve their score on the speech intelligibility test. IV. SUMMARY AND CONCLUSIONS This paper has presented the results of an experiment examining whether the decreases in localization ability that occur when listeners wear double hearing protection also impair their ability to segregate spatially separated speech signals. As in previous studies, the results of this experiment have shown that listeners who can reliably localize sound sources while wearing single hearing protection cannot do so Brungart et al.: Letters to the Editor 1899 when they are wearing double hearing protection. However, despite this decrease in localization ability, listeners wearing double hearing protection still seem to gain substantial, albeit somewhat reduced, intelligibility benefit from the spatial separation of the competing talkers. Thus it seems that the degraded localization cues that prevent listeners from hearing sound sources in the correct locations do not necessarily make all sound sources appear to originate from the same location. This suggests that listeners wearing double hearing protection may not be as impaired in their ability to process complex auditory scenes as their poor localization scores would suggest. It also offers some hope that listeners who have been thoroughly trained while wearing double hearing protection 共or those who are constantly exposed to boneconducted sound due to conductive hearing loss兲 may eventually be able to learn to improve their localization ability. Previous research has shown that listeners who receive training can learn to adapt to a variety of different disruptions in the cues they would normally use to localize sounds. For example, Hofman and his colleagues 共1998兲 conducted an experiment where listeners’ normal localization cues were disrupted by inserting plastic molds into their left and right ears. The results of this experiment showed that listeners could, over a several day period, learn to localize relatively well with these disrupted cues. Furthermore, once they adapted to these cues, they were then able to localize with or without the earmolds with no addional periods of adaptation. Shinn-Cunningham, Durlach, and Held 共1998兲 reported similar results for localization cues that were systematically remapped in the horizontal plane. Thus, to the extent that double hearing protection produces disrupted location cues that change systematically with source location, one might expect listeners to eventually be able to learn to use these disrupted cues to localize sounds. Further research is now needed to determine the extent, if any, to which listeners are also able to adapt to the disrupted localization cues that occur with double HP. ACKNOWLEDGMENT This work was sponsored in part by AFOSR Grant No. 01-HE-01-COR. This noise was generated by using a DSP processor 共Tucker-Davis RP2兲 to rectangularly gate the output of an analog pink noise generator 共GenRad兲. Note that, although the noise was both pink and broadband prior to transmission to the loudspeakers, no attempts were made to correct for the loudspeaker and amplifier responses to ensure that this signal was pink at the location of the center of the head. 1 1900 J. Acoust. Soc. Am., Vol. 116, No. 4, Pt. 1, October 2004 2 The standard deviations of these threshold measurements were approximately 5 dB in the No HP condition, 7 dB in the Single HP condition, and 13 dB in the Double HP condition. 3 Across the individual subjects in the Double HP condition, one listener performed relatively well at the left-right identification task 共⬇80% correct兲, one performed relatively poorly 共⬇30% correct兲, and the others fell between 50 and 70% correct responses. These differences are consistent with the relatively large individual differences found in our earlier study on localization with double hearing protection 共Brungart et al., 2003兲. Abel, S., and Armstrong, N. 共1993兲. ‘‘Sound localization with hearing protectors,’’ J. Otolaryngol. 22, 357–363. Abel, S., and Hay, V. 共1996兲. ‘‘Sound localization: The interaction of aging, hearing loss, and hearing protection,’’ Scand. Audiol. 25, 3–12. Berger, E. 共1983兲. ‘‘Laboratory attenuation of earmuffs and earplugs, both singly and in combination,’’ Am. Ind. Hyg. Assoc. J. 44, 321–329. Berger, E., Kieper, W., and Gauger, D. 共2003兲. ‘‘Hearing protection: Surpassing the limits to attenuation imposed by the bone-conduction pathways,’’ J. Acoust. Soc. Am. 114, 1955–1967. Bolia, R., D’Angelo, W., Mishler, P., and Morris, L. 共2001兲. ‘‘The effects of hearing protectors on auditory localization in azimuth and elevation,’’ Hum. Factors 43, 122–128. Bolia, R., Nelson, W., Ericson, M., and Simpson, B. 共2000兲. ‘‘A speech corpus for multitalker communications research,’’ J. Acoust. Soc. Am. 107, 1065–1066. Brungart, D., Kordik, A., Simpson, B., and McKinley, R. 共2003兲. ‘‘Auditory localization in the horizontal plane with single and double hearing protection,’’ Aviat., Space, Environ. Med. 65, A31–38. Cherry, E. 共1953兲. ‘‘Some experiments on the recognition of speech, with one and two ears,’’ J. Acoust. Soc. Am. 25, 975–979. Hofman, P. M., Van Riswick, J. G. A., and Van Opstal, A. 共1998兲. ‘‘Relearning sound localization with new ears,’’ Nat. Neurosci. 1, 417– 421. Levitt, H. 共1971兲. ‘‘Transformed up-down methods in psychoacoustics,’’ J. Acoust. Soc. Am. 49, 467– 477. Mills, A. 共1958兲. ‘‘On the minimum audible angle,’’ J. Acoust. Soc. Am. 30, 237–246. NIOSH 共1998兲. ‘‘Criteria for a recommended standard-occupational noise exposure,’’ Pub. No. 98 –126, U.S. Department of Health and Human Services, Washington, DC. Noble, W., Byrne, D., and LePage, B. 共1994兲. ‘‘Effects on sound localization of configuration and type of hearing impairment,’’ J. Acoust. Soc. Am. 95, 992–1005. Noble, W., Murray, N., and Waugh, R. 共1990兲. ‘‘The effect of various hearing protectors on sound localization in the horizontal and vertical planes,’’ Am. Ind. Hyg. Assoc. J. 51, 370–377. Shinn-Cunningham, B., Durlach, N., and Held, R. 共1998兲. ‘‘Adapting to supernormal auditory localization cues I: Bias and resolution,’’ J. Acoust. Soc. Am. 103, 3656 –3666. Simpson, B. D., Bolia, R., McKinley, R., and Brungart, D. 共2002兲. ‘‘Sound localization with hearing protectors: Performance and head motion analysis in a visual search task,’’ Proceedings of the Human Factors Society 46th Annual Meeting, Baltimore, MD, Sept 30–Oct 4, 2002, 100–104. Vause, N., and Grantham, D. 共1999兲. ‘‘Effects of earplugs and protective headgear on auditory localization ability in the horizontal plane,’’ Hum. Factors 41, 282–294. Zurek, P. 共1986兲. ‘‘Consequences of conductive auditory impairment for binaural hearing,’’ J. Acoust. Soc. Am. 80, 466 – 472. Zwislocki, J. 共1957兲. ‘‘In search of the bone conduction threshold in a free field,’’ J. Acoust. Soc. Am. 29, 795– 804. Brungart et al.: Letters to the Editor