| Article Access Statistics|
| Viewed||5327 |
| Printed||245 |
| Emailed||1 |
| PDF Downloaded||116 |
| Comments ||[Add] |
| Cited by others ||5 |
|Year : 2007
: 9 | Issue : 37 | Page
|Effects of binaural electronic hearing protectors on localization and response time to sounds in the horizontal plane
Eric L Carmichel, Frances P Harris, Brad H Story
Department of Speech, Language and Hearing Sciences University of Arizona, USA
Click here for correspondence address
The effects of electronic hearing protector devices (HPDs) on localization and response time (RT) to stimuli were assessed at six locations in the horizontal plane. The stimuli included a firearm loading, telephone ringing and .5-kHz and 4-kHz tonebursts presented during continuous traffic noise. Eight normally hearing adult listeners were evaluated under two conditions: (a) ears unoccluded; (b) ears occluded with one of three amplitude-sensitive sound transmission HPDs. All HPDs were found to affect localization, and performance was dependent on stimuli and location. Response time (RT) was less in the unoccluded condition than for any of the HPD conditions for the broadband stimuli. In the HPD conditions, RT to incorrect responses was significantly less than RT to correct responses for 120° and 240°, the two locations with the greatest number of errors. The RTs to incorrect responses were significantly greater than to correct responses for 60° and 300°, the two locations with the least number of errors. The HPDs assessed in this study did not preserve localization ability under most stimulus conditions.
Keywords: Electronic hearing protection, localization
|How to cite this article:|
Carmichel EL, Harris FP, Story BH. Effects of binaural electronic hearing protectors on localization and response time to sounds in the horizontal plane. Noise Health 2007;9:83-95
Hearing protection devices (HPDs) are used to reduce the potentially adverse effects of high-level noise on the human auditory system. They may be either passive such as circumaural earmuffs or earplugs or active, incorporating electronic circuitry that modifies incoming acoustic signals. Passive HPDs, especially earmuffs, are effective in attenua-ting hazardous noise levels but are known to alter the acoustic cues needed by a listener to localize sound sources ,,, and reduce speech understanding in noise.  These effects may occur even when listeners are able to use exploratory head movements to perform a localization task.  Electronic HPDs are purported by their manufacturers to alleviate some of these hazards (see Appendix), although these claims have not been substantiated. Determining the effects that electronic HPDs have on a wearer's ability to locate sound quickly and accurately is important for safety reasons, especially in industrial or outdoor recreational settings.
|How to cite this URL:|
Carmichel EL, Harris FP, Story BH. Effects of binaural electronic hearing protectors on localization and response time to sounds in the horizontal plane. Noise Health [serial online] 2007 [cited 2022 Oct 4];9:83-95. Available from: https://www.noiseandhealth.org/text.asp?2007/9/37/83/37424
Localization in the horizontal plane depends upon the listener's ability to use interaural time (ITD) and level (ILD) differences to judge a sound's location. ITDs dominate at frequencies below approximately 1700 Hz whereas ILDs provide cues above 1700 Hz. , Direction-dependent spectral cues are provided by the head and pinnae with the pinna providing cues for front/back discrimination above 3 kHz.  It is not clear whether these processes are altered when listeners are localizing sound sources in noisy environments. Localization ability may be degraded when the properties of a listener's head, pinnae and external auditory meatus are altered.
The circuitry of an electronic HPD is typically housed within a sound-attenuating circumaural earmuff, protective headgear or custom-fabricated earmoulds. When in place, they may obscure the pinna or alter the acoustical properties of the external auditory meatus and affect localization ability. The electronic components are designed primarily to reduce noise and not necessarily to enhance localization performance (c.f., Appendix). Two types of circuitry are used most commonly: (1) active noise reduction; (2) amplitude-sensitive sound transmission. The former type relies upon the principles of waveform superposition and destructive interference to cancel noise. Sound is detected inside the earcup by a microphone and its electrical output is fed back through an electronic processing circuit and amplifier and finally back to the earcup. The resultant signal is equal in amplitude but 180° out of phase with the original noise, thus causing energy cancellation.
Amplitude-sensitive sound transmission electronic earmuffs have one or more microphones working together with limiting amplifiers to transmit external sounds to earphones mounted inside the earcups. These electronic HPDs allow the user to hear low and moderate level sounds normally while providing attenuation for high-level noise. This is realizable with the use of a microphone, fast-acting automatic gain control (AGC) or peak-limiting circuitry and a miniature loudspeaker-receiver. Typically, the amplifier maintains a user-adjustable level until the external noise is great enough to limit the gain such that the only sound transmission is direct acoustic transmission through the earcup. Hearing protection devices with independent microphones on each side are further classified as binaural HPDs.
To date, a limited number of investigations of electronic HPDs and localization ability have been reported. These vary in the stimuli, loudspeaker arrangements and the methods used for device calibration, all of which may influence the outcome of a study. Abel and Armstrong  compared responses with an electronic HPD, earmuffs and earplugs to an unoccluded condition. Stimuli were 300-ms tonebursts at either .5-kHz or 4 kHz, presented in a background of 65 dB SPL white noise. Localization ability was degraded for the .5-kHz stimulus under all occluded conditions and there was no improvement with the electronic HPD. At 4 kHz, localization was poorest with the electronic HPD. The six loudspeaker locations did not include 0 and 180 degrees and the signals did not include broadband sounds that are more representative of naturally occurring sounds. Noble, Murray and Waugh  used a broader loudspeaker array and additional HPDs, all of which degraded localization ability, especially for front vs back positions. Only one signal was presented in quiet. A more comprehensive study by Abel and Paik  used three noise stimuli (.5- and 4-kHz NBN and BBN) to evaluate localization with automatic noise reduction devices. They determined that localization was degraded in all occluded conditions.
| Purpose and Aims|| |
This study was designed to expand on previous work by using a variety of stimuli delivered in an environment that simulated a real-world listening situation to investigate the effects of binaural electronic HPDs on localization. Specifically, we sought to determine if:
- A listener's ability to localize stimuli would be poorer when wearing electronic HPDs than when listening in an unoccluded condition.
- Types of localization errors would differ in occluded vs unoccluded conditions.
- Localization errors would vary as a function of stimulus type.
- Localization ability would vary as a function of the physical or electrical characteristics of the HPD.
- Response times would reveal information about judgment errors and listeners' perceived sound-source ambiguity.
Four male and four female listeners participated. They ranged in age from 20 to 40 years (mean = 24 years). All reported normal hearing, and this was confirmed by pure-tone audiometry at 125, 250, 500, 1000, 2000, 4000 and 8000 Hz for which threshold levels were less than 20 dB HL bilaterally. Interaural differences did not exceed 10 dB at any test frequency.
Four stimuli were used: two successive transient clicks of a semiautomatic handgun being loaded, the double ring of a telephone with a mechanical ringer and electronically generated frequency modulated (FM) tonebursts at .5 and 4 kHz. The first two stimuli were selected because they are broadband, complex and familiar. Both were digitally recorded in an anechoic environment so that "normal" reverberation and decay would occur as a consequence of the semireverberant test room and not from anomalous sources on the recording. Time domain and narrowband analyses of these two stimuli made from recordings acquired at the approximate position of the listener's head during the experiment, are shown in [Figure - 1]a and b.
As illustrated in the time-domain display in the leftmost panel of [Figure - 1]a, the duration of the firearm stimulus was approximately 1.1 s including decay time. The total stimulus was composed of a 120-ms duration "click" corresponding to the empty magazine being inserted into the handgun followed by 840 ms of silence and then a 90-ms click corresponding to the cocking of the handgun. The corresponding average spectrum of the stimulus is displayed in the corresponding rightmost panel. The amplitude of the stimulus's constituent frequencies was relatively uniform across frequencies, especially between 1 and 10 kHz. This stimulus was similar to the sound of cocking an M-16 assault rifle as used by Vause and Grantham.  The ability to localize this sound may be crucial to military and law enforcement personnel.
[Figure - 1]b illustrates the time- and frequency-domain recordings of the telephone-ringing stimulus. The duration of this stimulus including decay time was 2.5 s. The spectrum of the stimulus consisted of two fundamental frequencies at 1300 Hz and 1600 Hz with overtones as opposed to the broad-spectrum composition of the firearm loading.
The two electronically generated FM tones at .5 and 4 kHz were similar to stimuli used by Abel and Armstrong  and would not be frequently encountered sounds for most participants. Both were 300 ms in duration with 50-ms rise and fall times. These stimuli were included to investigate localization ability with HPDs as a function of frequency. Theoretically, dichotic electronic earmuffs should not alter the interaural time difference (ITD) used to localize the .5-kHz FM toneburst, and they should not appreciably alter the interaural level differences (ILD) used to locate the 4-kHz FM toneburst. Preservation of the ILD, however, assumes that the earmuffs have linear gain characteristics ( i.e ., no compression) at or below the stimulus presentation level. The brief duration (300 ms) of the tonebursts was chosen to test localization ability in a manner that might predict HPD performance in settings where rapid discrimination would be critical.
Background noise was presented continuously at 65 dB SPL throughout a test series. The noise was a looped, 4-channel recording of steady traffic noise captured in the soundfield. This noise was selected instead of the broadband noise used in other studies , because it is more representative of noise that might occur in a "real-world" environment. A fast Fourier transform (FFT) of the traffic noise is shown in [Figure - 2]. For comparison, a 65 dB SPL white noise signal is also plotted. The dominant low-frequency characteristic of the traffic noise with a rolloff of energy above approximately 2 kHz is evident.
Test environment including loudspeaker arrangement and calibration
Listening tests were conducted in a semireverberant room measuring 4.88 m long, 4.27 m wide and 2.74 m high. Loudspeaker placement was optimized to assess the types of localization errors (contralateral or ipsilateral) while maintaining a reasonable length of time for the experiment. Six loudspeakers positioned 60 degrees apart at azimuth angles of 0°, 60°, 120°, 180°, 240° and 300° and a distance 1.83 m from the center listening position formed a circular array. Four additional loudspeakers spaced 90 degrees apart and 2.60 m from the listening position were used to provide background noise. The loudspeakers were arranged not only for symmetry but also to maximize their distance from the listener. The greater the distance from the listener to the sound source, the less likely that changes in head position could affect perceived loudness or the measured intensity at the head position.
The ten loudspeakers were RCA/Radio Shack catalog #40-5007 full-range speakers. As multi-track recordings were used to present stimuli and background noise, each loudspeaker had a dedicated power amplifier. The gain of each amplifier was adjusted so that sound levels produced from each loudspeaker were within 0.5 dB SPL for frequencies .5, 1 and 2 kHz. The height (floor to under surface of loudspeaker) for each loudspeaker was 1.02 m, which put the approximate acoustic centers of the loudspeakers at 1.22 m above ground. A height adjustable chair was used to place participants' ears (approximate center of conchas) at 1.22 m above ground level.
Calibration of the loudspeakers was performed using a 1-inch Brüel and Kjaer microphone equipped with a random incidence corrector. The microphone and its respective preamplifier were connected to a real-time analyzer (RTA). For calibration purposes, all sound pressure levels were measured at a point representing the listener's head position: The microphone was positioned such that its axis was parallel to and 1.22 m above the floor. The front edge of the microphone coincided with a vertical line extending through the center point of the loudspeaker circular array. Although a random incidence corrector was used, the microphone was continually repositioned to maintain an angle of incidence equal to 30 degrees for each loudspeaker undergoing calibration.
Stimulus delivery and response box
A personal computer with a MOTU 828 FireWire audio interface, a Roland ED UA-30 USB audio interface and Cool Edit Pro multitrack recording software were used to present the stimuli and background noise. Custom-built hardware and software controlled the signal presentation and aligned response times (RTs) with the system's internal clock. A hand-held response box was constructed with six push-button switches that were arranged to correspond with the azimuth angles of the loudspeakers. Listeners indicated their choice by pressing the button on the response box corresponding to the sound's perceived direction. Associated data-logging hardware was used to record listeners' response choices (azimuth angle) and RT. The response box was electrically connected to a microprocessor-controlled receiver that in turn provided optical isolation from the computer.
Hearing protection devices
Three commercially available HPDs with unique physical and electrical designs that represented the industry standards were included. These are described in detail in the Appendix and are referred to in the remainder of the article as: HPD-A, R2000 Electronic Thin Muff; HPD-B, Pro-Ears Dimension 2; HPD-C, Action Ear Sport. All were dichotic, amplitude-sensitive sound transmission earmuffs and were purchased "off-the-shelf" to ensure a valid representation of commercial products.
The three electronic HPDs were physically and electrically different from each other. It was hypothesized that localization ability for each HPD could be a function of the earmuff's physical design ( e.g ., microphone orientation or earcup shape) as well as its electrical characteristics such as frequency response or distortion. Notable differences in physical design included: (1) Microphones for HPDs A and B are located on the surface of the earcups whereas microphones for HPD-C are separate from the earcups and are located proximal to the wearer's temples (which purportedly enhances localization); (2) HPD-A has thin profile earcups that place the microphones proximal to the wearer's head when compared to the deeper earcups of HPD-B; (3) the microphones of HPD-A lie in the anatomical frontal plane whereas those of HPD-B are slightly anterior of the frontal plane. (Note: according to the manufacturer, these HPDs can be worn with the microphones directed towards the front or back.)
HPD calibration and evaluation
Calibration of the HPDs is an important consideration when determining localization responses. The interaural level difference required to detect a change in position from a fused image at midline (the perceived location under earphones when there is no interaural level difference) is from 0.5 to 1.0 dB for frequencies between 200 and 5000 Hz.  Differences between earcup levels greater than 1.0 dB could go unnoticed without a sensitive method of measuring HPD acoustical gain. Noble, Murray and Waugh  balanced the gain settings of the HPD earcups by averaging the judgments determined empirically by the three authors. We used an objective method for adjusting the acoustical gain of each earcup of the HPDs prior to their evaluation.
The gain response of an electronic HPD is nonlinear above a predetermined input level. Both this kneepoint and the types of gain reduction above this level are proprietary for each manufacturer. Electro-acoustical measurements were performed and indicated that all of the earmuffs had linear I/O characteristics with no compression or automatic gain control (AGC) activity at and below 75 dB SPL for 2-kHz tonebursts. Therefore each stimulus was presented at 72 dB SPL during testing as measured at the listener's ear level at the center of the loudspeaker array.
All of the HPDs had independent gain controls for each earcup. A 2-kHz signal at 72 dB SPL was chosen to set the acoustical input/output ratio of each earcup to unity or 0 dB. Setting the gain too high could result in discomfort or in potential damage to the hearing of the listener. Having the gain too low would preclude the usefulness of binaural electronic HPDs. Differences in gain between left and right earcups would result in differences in intensity at each ear, thus compromising localization ability. All earcups for the three HPDs were adjusted by the experimenter to deliver 72 dB SPL at the listener's ear when using a 2-kHz, 72 dB SPL input signal (this acoustical input/output ratio does not equal one at all frequencies or for higher-intensity input levels). The 2-kHz frequency was selected because HPDs have greater passive attenuation at high frequencies versus low frequencies. Consequently, a high-frequency input signal was more readily isolated from the output signal when a tight acoustic seal was achieved. Without input/output isolation, it would be difficult to determine if the measured output level was energy emitted by the earphone or acoustical energy in the sound field. It was not feasible to use a frequency higher than 2 kHz because the HPDs have a limited response to high frequencies.
Calibrating the electronic HPDs required further consideration because the input signal was acoustical energy in the soundfield. The only isolation of the input signal from the output signal was the passive attenuation provided by the HPDs themselves. Attempts to place the HPD earcups on standard audiometric couplers used for audiometric earphones ( e.g ., Brüel and Kjaer artificial ear type 4152 used for calibrating Telephonics "TDH" earphones) did not provide sufficient input-to-output acoustical isolation because of the shape of the HPD earcup cushions. Similar difficulties were encountered when using the Knowles Electronic Manikin for Acoustical Research (KEMAR). Therefore, a specialized HPD calibration and assessment system was developed-the Profile Head for Acoustical Noise Testing and Otological Measurements (PHANTOM). The PHANTOM is a 16 cm × 13 cm × 28 cm rectangular assembly [Figure - 3]. The advantage of the PHANTOM is that it has two matched "ears" per side of its "head", which allowed simultaneous real-time comparisons of an unoccluded vs protected ear. Each PHANTOM "ear" was cast from an average-sized human ear and it consisted of a concha plus canal coupled to a ¼-inch condenser microphone. Circular and elliptical markings on each side facilitated centering the HPDs over the microphones. A strain gauge built into the PHANTOM allowed measurement of HPD headband force. No leakage of sound from around the earcups of the HPDs or absorption by the materials used to construct the PHANTOM was measured, eliminating these factors as sources of error.
The frequency responses obtained for the HPDs placed on the PHANTOM are shown in [Figure - 4]a-c. These measurements were made in the semireverberant test room with the PHANTOM placed at the center of the loudspeaker array and with its "ears" positioned at the theoretical level of a listener's ears. Loudspeakers located at azimuth angles of 60° and 300° produced the swept tones used for making the right and left earcup measurements, respectively. A constant SPL was maintained between 0.2 and 10 kHz by use of a Brüel and Kjaer compression amplifier and a ½-inch Brüel and Kjaer reference microphone. HPD-A and HPD-B had a relatively uniform response across the frequency range as seen in [Figure - 4]a and b. The factory modification of HPD-C resulted in an emphasis in the frequency response below 2 kHz as seen in [Figure - 4]c. Thus, the background noise and the 0.5-kHz toneburst were enhanced when using this HPD in comparison to the unoccluded condition.
Microphones internal to the PHANTOM were calibrated using three different methods to ensure valid measurements. First, the microphones in the PHANTOM were removed and calibrated using a Brüel and Kjaer Type 4230 acoustic calibrator with a ¼-inch adapter. This provided accurate calibration of the microphones at 1 kHz outside of the PHANTOM but it excluded the response peak imparted by the artificial head's conchas. Second, the microphones were evaluated for response uniformity without removing them from the PHANTOM by using Sennheiser HDA 200 earphones, which were factory-calibrated using a Brüel and Kjaer 4153 artificial ear with standard cone (YJ0304) above an adapter plate (Type DB 0843). The Sennheiser headphones were fit on the PHANTOM in the same fashion as the HPDs. They provided 72 dB SPL output from 0.1 to 10 kHz using a 150 mV (nominal) swept sinewave signal. Finally, sound-field calibration of the microphones was accomplished using a Tektronix FG 507 sweeping function generator, a Brüel and Kjaer compression amplifier, a full-range loudspeaker with power amplifier and a calibrated ½-inch Brüel and Kjaer microphone. A constant SPL from the loudspeaker was maintained across test frequencies at the listening position by use of the compression amplifier and calibrated microphone. The microphones inside the PHANTOM were connected to a real-time FFT analyzer to determine their sensitivity and response characteristics. As the profile head's microphones were acoustically coupled to the sound field via artificial ear canals and conchas, a peak in the frequency response due to resonance was measured and anticipated. At 0.5, 1 and 2 kHz, which are below the resonance peak, all of the above calibration schemes yielded identical output levels from the microphones.
| Procedure|| |
Participants were seated comfortably in the center of the loudspeaker array and were instructed to position and maintain their heads directly facing the loudspeaker at 0° azimuth during stimulus presentation. Two LEDs on that loudspeaker blinked prior to each sound presentation, thus alerting the listener that a sound would be presented in 1 s. The LEDs' location also provided a convenient visual target for maintaining focus straight ahead. The stimuli were presented individually with a 10-s inter-stimulus interval. Listeners were informed that the presentation order of stimuli and locations were random, but they were not informed that each stimulus would be presented an equal number of times per loudspeaker. Listeners used the handheld response box to indicate their selection of sound location. They were instructed to guess if they were unsure, and response feedback was not provided. A forced-selection paradigm was used in which the system paused the recording automatically if the listener did not make a response within 5 s of a stimulus presentation. The system was configured to begin or resume normal play mode by depressing any of the response box's switches. "Markers" on the recording ensured stimulus-RT synchronicity even when the recording paused or stopped during a test. The recorded information made it possible to observe errors and RT as a function of stimuli, azimuth angle, participant, test sequence or HPD. Each RT measurement was relative to the onset of its respective stimulus.
A 4-min trial presentation that included all four stimuli was administered in an unoccluded condition prior to beginning the test. For this practice test, each stimulus was presented once per loudspeaker at all locations for a total of 24 stimuli. The presentation order of the stimulus-loudspeaker combinations was random. Upon completion of the trial, listeners were tested in the unoccluded condition (Condition 1) and in three occluded conditions using the three electronic HPDs (Conditions 2, 3 and 4). The presentation order of the four conditions was counterbalanced across listeners. The investigator fit each HPD. The controls for each HPD were preset to provide an acoustical I/O ratio of one (or unity acoustical gain) and remained fixed at this level for the duration of the experiment. Calibration checks were performed between every other test session to ensure consistent amplification for all HPDs. None of the HPDs required readjustment subsequent to the initial calibration. Each stimulus was presented a total of three times per loudspeaker at all azimuth angles for one test block. Each participant completed four listening blocks, one unoccluded and three occluded conditions where one "block" consisted of 72 stimuli (4 stimuli × 3 trials × 6 loudspeakers). The presentation order of the stimulus-loudspeaker combinations was random. Testing for each participant was divided into two sessions (two blocks per session) to minimize fatigue. The first session for each participant required approximately 50 min and included instructions, the trial presentation and two 12-min blocks. The second session, which occurred on a different day, consisted of two 12-min blocks. No participants required time for reorientation to the task.
| Data analysis|| |
Statistical analysis was performed using a general linear model (GLM) analysis of variance (ANOVA) for mixed effects (model III) and multifactor crossed classification experiments. A type III sum of squares was used in all analyses. A separate analysis was performed for each of the four stimuli. The two dependent variables, localization scores and RT were analyzed separately. As the system was designed to pause the recording automatically if the listener did not make a response within 5 seconds of a stimulus presentation, RTs ≥ 5 seconds were excluded from the analysis. The total number of responses was 2304 out of which, 2272 were included in the RT analysis. All localization responses were considered valid because listeners were instructed to guess if they were unsure. Response time to incorrect and correct judgments for each stimulus and condition were also compared. The four independent variables or factors and their respective levels are listed in [Table - 1]. Both SAS (version 8.0) and SPSS (version 11.0) analysis software were used for data analysis.
| Results|| |
The mean localization scores as a function of condition (unoccluded vs HPDs) for each stimulus are plotted in [Figure - 5]. Differences in localization among conditions were significant for the firearm, F (3,21) = 15.30, P < .05, telephone, F (3,21) = 15.40, P < .05 and 4-kHz toneburst, F (3,21) = 10.80, P < .05 stimuli, but not for the 0.5-kHz toneburst, F (3,21) = 1.13, P = .36. Fewer localization errors were made in the unoccluded condition than in any of the HPD conditions for all stimuli. Results varied by stimulus type and azimuth angle with HPDs. The broadband stimuli (firearm loading and telephone ring) were the most accurately localized. The mean correct responses (averaged across all azimuth angles) for localization to broadband stimuli in the unoccluded condition were 97.2 (firearm) and 98.6% (telephone). Mean scores for localization to these stimuli ranged from 67.4-54.2% correct in the HPD conditions when averaged across all locations. There was a stimulus-location interaction, F (15,105) = 6.96, P < 0.01, when the data were analyzed across all conditions, locations, stimuli and participants.
[Figure - 6] represents localization accuracy as a function of condition, stimulus and azimuth angle. Localization accuracy with HPDs was best for azimuth angles of 60° and 300° for the broadband stimuli as seen in [Figure - 6]a and b. When averaged across these locations, the mean correct responses for localization to broadband stimuli for HPDs ranged from 79.2-87.5% correct.
The mean localization scores for each condition are displayed in [Figure - 6]c and d as a function of location for the .5 and 4-kHz stimuli. The .5-kHz stimulus was difficult to localize and there was a high incidence of errors regardless of condition. Mean localization scores in the unoccluded condition (45.8% correct when averaged across all locations) and the HPD conditions (scores ranged from 42.4-37.5% correct when averaged across all locations) were similar. There was no difference in localization accuracy among conditions, F (3,21) = 1.13, P = 0.36. Mean scores for localization to the 4-kHz stimulus (averaged across all locations) ranged from 59.0% correct (unoccluded condition) to 27.8% correct (HPD-C). Localization errors (among all conditions) for the 4-kHz toneburst were significant (F (3,21) = 10.77, P < 0.05), HPD-A and HPD-B allowed better localization than did HPD-C. However, the lowest mean scores at 240º (0% correct) and 120º (8.3% correct) were obtained with HPD-A.
The sources of listeners' errors are delineated for both the unoccluded and HPD conditions for the broadband stimuli in [Table - 2]. The results from the HPDs were combined for this summary because there were no significant differences between them for these stimuli. Of the total of 1152 responses, there were no left-right confusions for the broadband stimuli for any condition.Front-back reversals accounted for the largest percentage of errors in the HPD conditions.
| Response time|| |
The mean RTs as a function of condition for each stimulus are displayed in [Figure - 7]. There were differences among the conditions for the broadband stimuli (firearm, F (3,21) = 6.15, P < 0.05 and telephone, F (3, 21) = 5.21, P < 0.05). The RT for the unoccluded condition was less than the HPD conditions when averaged across all locations. There were differences among the HPD conditions. Mean RT for HPD-B was less than for HPDs A and C for the firearm stimulus. Mean RTs for HPD-A and HPD-B were less than HPD-C for the telephone stimulus. There were no differences in RT among conditions for the .5-kHz (F (3,21) = 2.74, P = 0.07) or the 4-kHz stimuli (F (3,21) = 1.86, P = 0.17). Response times to correct vs incorrect responses were analyzed to determine the confidence of the listeners' judgments. Descriptive statistics were used to characterize the effects of condition, location and stimulus on RT and localization accuracy. When averaged across all stimuli and locations, the mean RT to correct responses in the unoccluded condition (M = 1.58 s, SD = 0.76) was not different from the mean RT to errors (M = 1.64 s, SD = 0.81). In the HPD conditions, the mean RT to incorrect responses was less (M = 1.88s, SD = 0.89) than to correct responses (M = 1.90s, SD = 0.81) but the differences were not significant.
[Figure - 8] illustrates that mean RT as a function of accuracy (correct vs incorrect responses) varied with location. In the HPD conditions, RT to incorrect responses (M = 2.07 s, SD = 0.74) was significantly less ( t (282), P < 0.01) than RT to correct responses (M = 2.46 s, SD = 0.64) for 120° and 240°, which were the locations with the greatest number of errors. On the other hand, the RT to incorrect responses (M = 2.59 s, SD = 0.79) was significantly greater ( t (284), P < 0.01) than to correct responses (M = 2.07 s, SD = 0.71) for 60° and 300°, which were the locations with the least number of errors. The mean RTs plotted in [Figure - 8] were averaged across the firearm and telephone stimuli, which were the easiest stimuli to localize, but the trend remained unchanged.
| Summary and Discussion|| |
The purpose of this study was to investigate the effects of electronic HPDs with different physical and electrical designs on localization ability and response times to stimuli in the horizontal plane. Listening tests were performed in a semireverberant room using stimuli of different spectral content, duration and "familiarity" in a background of traffic noise to approximate real-world environments more closely. Response time needed to localize stimuli was included as a measure of listeners' uncertainty (or task difficulty). There were differences in localization between the unoccluded and HPD conditions for the firearm, telephone and 4-kHz stimuli but not for the .5-kHz stimulus. This stimulus was difficult to localize and there was a high incidence of errors regardless of condition. Based on these findings, it is not possible to determine if the HPDs had an effect on the ITD cue used for localizing low-frequency (<1700 Hz) sounds. Despite physical design and electro-acoustical performance differences among the HPDs tested, there was only one difference between them in listeners' ability to localize sounds: Use of HPD-A resulted in more correct decisions than did HPD-C for the 4-kHz stimulus when scores were averaged across all locations. Differences in frequency response or intrinsic noise may have been responsible for the effect. HPD-C had decreased sensitivity to 4-kHz and a degraded signal-to-noise ratio due to its low-pass filter (see Appendix). With regard to physical design differences, the microphones of HPD-C are placed in a unique position in front of the earcups. It cannot be determined if this offered any advantage or disadvantage because other performance variables could have imparted a greater effect on localization scores.
The choice of stimuli is an important consideration when evaluating HPDs for localization performance. Low- and high-frequency FM tonebursts were included to investigate the extent to which ILD and ITD localization cues are affected by electronic HPDs. Localization of the .5 kHz toneburst was poor in any condition regardless of azimuth angle whereas for the 4-kHz toneburst, localization was angle-dependent. When results for this stimulus were averaged across 120°, 180° and 240° locations - sites with the greatest number of errors -there was no difference among conditions and HPD-A had the lowest score at 240° (0% correct). Based on these results and the stimulus-dependent nature of the errors, assessment of HPDs in noise and semireverberant conditions using electronically generated tonebursts is not considered to be a reliable method. Stimulus duration could have affected performance for the FM tones. The duration of the tonebursts (300 ms including 50 ms rise and fall times) was less than the telephone's ring (approximately 2.5 s) but greater than the 0.12 s and 0.09 s clicks for the firearm's loading. The telephone used for making the recording provided an interesting stimulus because its fundamental frequencies were at 1300 Hz and 1600 Hz, which are close to the approximate ITD/ILD "break" frequency of 1700 Hz. The click of the firearm's loading resulted in a broad-spectrum noise. Despite these considerable differences in spectral content and duration of the broadband stimuli, there were no differences in localization accuracy between them.
For all conditions, more ipsilateral judgment errors were made to sounds coming from the rear (180° azimuth) than for sounds coming from the front (0° azimuth). Previous research  has shown that earplugs have the reverse effect and it was believed that loss of canal resonance accounted for the rearward illusion. Frequency response curves measured from the PHANTOM revealed a resonant peak commensurate with the normal ear canal resonance under earmuffs in both active [Figure - 4] and passive modes of operation. Presence of this resonance peak may have accounted, in part, for listeners' propensity to a frontal plane illusion. Most errors made for the HPD conditions occurred at 120° and 240° [Figure - 7] and sounds coming from these locations were often judged as coming from 60° and 300°, respectively. Front-back reversals accounted for the largest percentage of the errors. One listener, however, made localization errors opposite from the other listeners. For this listener, regardless of condition, more ipsilateral errors were made to sounds coming from 0° than for sounds coming from 180°. Localization under HPDs for this listener was also unique: Stimuli presented at 60° and 300° were often judged to originate from 120° and 240°, respectively, which was opposite from the other listeners. This listener made 38.9% errors at 60° and 300° when averaged across all HPD conditions for the broadband stimuli, compared to 16.3% errors for the same factors when averaged across all listeners. Listening under HPDs did not change the rearward plane illusion for this listener, which suggests that preservation of the ear canal resonance under earmuffs retains a listener's natural (unoccluded) inclination to a forward or rearward judgment.
Results from this study suggest that localization ability with electronic HPDs is akin to lateralization with stereo headphones: It is easy to discern left-from-right sound source location but discrimination between left rear and left front (or right rear and right front) is difficult. Such errors were anticipated despite the "stereo" sound provided by the HPDs because the ITD of sound to the tympanic membrane does not uniquely specify a location in space, only the right/left component.  The ILD of sound is theoretically retained with binaural electronic HPDs, but the cues normally provided by the pinnae are obscured by the earcups. Results from this investigation suggest that binaural electronic HPDs retain the ILD cue needed for lateralization, assuming that the earcups are properly adjusted to give equal acoustical gain at each ear. However, pinna-head cues needed to make accurate front/back judgments are not retained.
Unoccluded localization ability was significantly poorer for the .5-kHz stimulus than for broadband stimuli and suggests that ITD does not provide the dominant cue used to locate broadband stimuli, at least in noisy and reverberant conditions. It is not possible to determine if background noise or the semireverberant environment made the .5-kHz stimulus difficult to localize. The background noise used in this investigation had greater low-frequency content [Figure - 2] thus providing more low-frequency masking than the white-noise maskers used by Abel and Hay  or Lorenzi, Gatehouse and Lever.  Considering that a 10-dB signal-to-noise ratio has an effect on localization for low frequencies when a white-noise masker is used, , it is probable that the traffic noise did impact localization of the 0.5-kHz stimulus.
With regard to the 4-kHz stimulus, the types of errors made in the unoccluded condition were similar to the errors made under HPDs for broadband stimuli. In the unoccluded condition, localization ability was poor (77% errors) for azimuth angles of 120 and 240°. Even so, there were no contralateral reversals for the 4-kHz stimulus in this condition. Therefore, cues used to make front/back localization judgments do not provide sufficient location-specific information for pure-tone sounds because the relative spectrum remains unchanged regardless of the sound's direction. For broadband sounds, however, location would have a greater effect. In this case, intensity enhancement resulting from canal resonance and attenuation imparted by the head-pinna interaction would modify the relative levels of a sound's component frequencies as a function of direction. Localization ability with HPDs and for azimuth angles of 120 and 240° was poor (64% errors when averaged across all HPDs). However, there were no contralateral reversals for the broadband stimuli in the HPD conditions. Therefore, the ILD cue is preserved and only the pinna-head cue is disrupted with binaural electronic HPDs. It is speculated that the pinna-head transfer function provides the dominant cue needed to localize broadband sounds in a noisy and reverberant environment.
There were significant differences in RT between correct and incorrect responses for the telephone ring and the 4-kHz toneburst, with longer RT to incorrect than to correct responses. Response time as a function of accuracy (correct vs incorrect responses) was dependent on condition and location. When averaged across all stimuli and locations, the mean RT to errors was almost equal to the RT to correct responses for the HPD conditions. The difference between mean RT to correct v s incorrect responses under HPDs was less than 20 ms and was not significant indicating that sound-source location was not ambiguous to listeners. Location had an interesting effect in the HPD conditions.
The mean RT for errors at 60 and 300° azimuth, which were locations with the least number of errors, was greater than the mean RT to correct judgments. The reverse occurred for 120 and 240° azimuth, which were locations with the greatest number of errors: The mean RT to errors was less than to correct judgments. This suggests further that the perceived sound-source location was not ambiguous.
Despite manufacturers' claims, results from this study are consistent with other studies , in showing that localization ability is compromised with binaural electronic HPDs. Localization performance may have improved by permitting head movement. Noble  measured better localization performance under passive HPDs when listeners were free to move their heads. However, the RTs that we measured in the HPD conditions suggest that for some locations, a sound's direction is not ambiguous, albeit judged incorrectly and additional time or head movement would not likely have enhanced localization.
| Conclusions|| |
Our results suggest that binaural electronic HPDs should be used with caution in occupational or recreational environments where determining a sound's direction is critical. However, if localization ability is not of concern, then electronic HPDs may enhance communication and safety by allowing the wearer to hear sounds that passive HPDs would otherwise attenuate. All of the HPDs used in this study had gain controls for each earcup that a listener could adjust for comfort. For this investigation, the HPD gain controls were carefully adjusted, and there were few contralateral reversals.
Random adjustment of the gain controls, however, could result in greater and more severe localization errors. It may not be obvious to inexperienced listeners when the gain between earcups is not equal. Additionally, individuals with hearing loss may experience other complications when adjusting gain. Individual dexterity and the ability to manipulate the controls should also be considered. At the very least, individuals should be informed that all HPDs have limitations and the limitations of any specific model of HPD should be understood before it is worn in hazardous conditions.
| Acknowledgements|| |
Portions of this study were presented at the American Academy of Audiology 16 th Annual Convention and Expo, 2004, Salt Lake City, Utah.
| References|| |
|1.||Abel SM, Armstrong NM. Sound localization with hearing protectors. J Otolaryngol 1993;22:357-63. [PUBMED] |
|2.||Abel SM, Hay VH. Sound localization: The interaction of aging, hearing loss and hearing protection. Scand Audiol 1996;25:3-12. [PUBMED] |
|3.||Abel SM, Kunov H, Pichora-Fuller MK, Alberti PW. Signal detection in industrial noise: Effects of noise exposure history, hearing loss and the use of ear protection. Scand Audiol 1985;14:161-73. [PUBMED] |
|4.||Atherley GR, Else D. Effect of ear-muffs on the localization of sound under reverberant conditions. Proc Roy Soc Med 1971;64:203-5. [PUBMED] [FULLTEXT]|
|5.||Abel SM, Alberti PW, Haythornthwaite C, Riko K. Speech intelligibility in noise: Effects of fluency and hearing protector type. J Acoust Soc Am 1982;71:708-15. [PUBMED] [FULLTEXT]|
|6.||Nobel W, Murray N, Waugh R. The effect of various hearing protectors on sound localization in the horizontal and vertical planes. Am Ind Hyg Assoc J 1990;51:370-7. |
|7.||Mills AW. Auditory localization. In : Tobias JV, editor. Foundations of modern auditory theory, vol. 2. Academic: New York; 1972. p. 301-48. |
|8.||Yost WA. Discrimination of interaural phase differences. J Acoust Soc Am 1974;55:1299-303. [PUBMED] [FULLTEXT]|
|9.||Musicant AD, Butler RA. The influence of pinnae-based spectral cues on sound localization. J Acoust Soc Am 1984;75:1195-200. [PUBMED] [FULLTEXT]|
|10.||Abel SM, Paik JE. Sound source identification with ANR earmuffs. Noise Health 2005;7:1-10. |
|11.||Vause NL, Grantham DW. Effects of earplugs and protective headgear on auditory localization ability in the horizontal plane. Hum Factors 1999;41:282-94. [PUBMED] |
|12.||Lorenzi C, Gatehouse S, Lever C. Sound localization in noise in normal-hearing listeners. J Acoust Soc Am 1999;105:1810-20. |
|13.||Mills AW. Lateralization of high-frequency tones. J Acoust Soc Am 1960;32:132-4. |
|14.||Russell G, Noble WG. Localization response certainty in normal and in disrupted listening conditions: Towards a new theory of localization. J Aud Res 1976;16:143-50. |
|15.||Oldfield SR, Parker SP. Acuity of sound localisation: A topography of auditory space: II, Pinna cues absent. Perception 1984;13:601-17. [PUBMED] |
|16.||Noble WG. Earmuffs, exploratory head movements and horizontal and vertical sound localization. J Aud Res 1981;21:1-12. [PUBMED] |
Brad H Story
1131 E. 2nd St. P.O. Box 210071, Tucson, AZ 85721
Source of Support: None, Conflict of Interest: None
[Figure - 1], [Figure - 2], [Figure - 3], [Figure - 4], [Figure - 5], [Figure - 6], [Figure - 7], [Figure - 8]
[Table - 1], [Table - 2]
|This article has been cited by|
||The Effectiveness of Hearing Protection Devices: A Systematic Review and Meta-Analysis
| ||Chanbeom Kwak, Woojae Han |
| ||International Journal of Environmental Research and Public Health. 2021; 18(21): 11693 |
|[Pubmed] | [DOI]|
||Noise and communication: A three-year update
| ||Brammer, A.J. and Laroche, C. |
| ||Noise and Health. 2012; 14(61): 281-286 |
||A multidimensional evaluation of auditory performance in one powered electronic level-dependent hearing protector
| ||Giguère, C., Laroche, C., Vaillancourt, V., Shmigol, E., Vaillancourt, T., Chiasson, J., Rozon-Gauthier, V. |
| ||19th International Congress on Sound and Vibration. 2012; : 1660-1667 |
||Advanced hearing protection and communication: Progress and challenges
| ||Giguére, C., Laroche, C., Brammer, A.J., Vaillancourt, V., Yu, G. |
| ||Proceedings of the Institute of Acoustics. 2011; 33 1((PART 3)): 225-233 |
||Sound localization with an army helmet worn in combination with an in-ear advanced communications system
| ||Abel, S.M., Boyne, S., Roesler-Mulroney, H. |
| ||Noise and Health. 2009; 11(45): 199-205 |