Field Trial Evaluations of a Switched Directional/Omnidirectional In-the-Ear Hearing Instrument

Transcription

1 J Am Acad Audiol 10 : (1999) Field Trial Evaluations of a Switched Directional/ In-the-Ear Hearing Instrument David A. Preves* Carol A. Sammeth Michael K. Wynne` Abstract The use of directional microphones is one of the few methods available for hearing aids to increase the signal-to-noise ratio. The smaller microphones available with today's technology have increased the viability of their application for in-the-ear (ITE) hearing aids. This study evaluated an ITE hearing aid containing two nondirectional microphones that provides wearer-selectable omnidirectional/directional operating modes. Ten sensorineural hearingimpaired patients were fitted binaurally. During the first trial period, the low-frequency gain decrease produced by the directional mode was not compensated for. The frequency responses were matched during the second trial period. For both trial periods, Hearing in Noise Test results using two uncorrelated noise sources indicated significant speech recognition improvements for the directional mode relative to the omnidirectional mode. Responses on Abbreviated Profile of Hearing Aid Benefit questionnaires, paired-comparison judgments, and interview data revealed that most subjects preferred the directional mode in noisier environments, but many also preferred the omnidirectional mode in quiet listening. Key Words : Amplification, directional microphones, hearing aids, speech recognition Abbreviations : AGC-O = automatic gain control with output compression, APHAB = Abbreviated Profile of Hearing Aid Benefit, BTE = behind-the-ear hearing aid, CD = compact disc, DI = Directivity Index, HINT = Hearing in Noise Test, ITE = in-the-ear hearing aid, MCL = most comfortable listening level, NAL-R = National Acoustics Laboratories' revised prescriptive formula, SNR = signal-to-noise ratio, UCL = uncomfortable loudness level he use of directional microphones is one of the few methods available to increase T the signal-to-noise ratio (SNR) for a hear- ing instrument. The advantages of directional microphones in behind-the-ear (BTE) hearing instruments have been well documented. For example, Lentz (1972) found 16 percent and 24 percent improvements in speech discrimination scores at test SNRs of 0 and -6 db, respectively, for 20 hearing-impaired persons wearing a directional BTE hearing aid as compared to a non- *Micro-Tech Hearing Instruments, Minneapolis, Minnesota, troudebush VA Medical Center, Indianapolis, Indiana,currently Otologics, LLC, Bolder, Colorado, tdepartment of Otolaryngology, Head and Neck Surgery, Indiana University School of Medicine, Indianapolis, Indiana Reprint requests : David A. Preves, Micro-Tech, Inc., P.O. Box 59124, Minneapolis, MN directional (omnidirectional) BTE hearing aid. Nielsen (1973) compared speech discrimination scores in four different SNRs and subjective preferences in simulated group conversation for a directional microphone and a nondirectional microphone in BTE hearing aids fitted binaurally on experienced hearing aid wearers. He found that discrimination scores were significantly better with the directional microphone at SNRs of +5 and +10 db but were not significantly different at SNRs of +15 db and +20 db. Fifteen of the 22 subjects preferred the directional microphone to the nondirectional microphone in a group conversation simulation. When directional hearing aids were first introduced in the 1970s, hearing aid designers and investigators soon realized that having a fixed amount of directionality was not optimal for all listening situations. It was desirable to give the wearers control of the amount of directivity provided by the hearing aids. For example, 273

2 Journal of the American Academy of Audiology/Volume 10, Number 5, May 1999 Nielsen (1973) suggested that this could be achieved by providing switchable networks that would allow a wearer-selectable choice of directional or omnidirectional functioning. Thus, incorporating a variable amount of directionality in BTE hearing aids is not a new idea. Early versions of BTE hearing aids that had variable amounts of directivity used mechanical shutters or valves to continuously adjust the amount of directionality (e.g., Knowles and Carlson, 1973 ; Gage and Muttick, 1974 ; Johanson and Stutz, 1974 ; Cole, 1977). In one evaluation of this technology, Mueller et al (1983) found that after 6 weeks, most of 30 hearing-impaired subjects (who had never worn hearing aids previously) preferred the directional to the omnidirectional mode on their switchable BTE instruments in noisy listening situations. Most recently, BTE hearing aids with switched directional/omnidirectional modes have been implemented with two microphones rather than one microphone with two sound inlets. Using a programmable directional/omnidirectional BTE hearing aid with two microphones, Valente et al (1995) found a mean improvement in SNR of 7 to 8 db with the Hearing in Noise Test (HINT) (Nilsson et al, 1994) for 50 experienced hearing aid wearers at two sites for the directional mode in comparison to the omnidirectional mode. In this instrument, the wearers controlled whether the hearing aid operated in directional or nondirectional mode via a remote control. Some hearing health care professionals believed that, until recently, directional microphones have been available only in BTE hearing instruments (e.g., Chasin, 1994). On the contrary, review of the literature reveals that directional microphones have also been available in in-the-ear (ITE) hearing instruments for many years (e.g., Preves, 1976 ; Rumoshovsky, 1977). However, since their introduction over 20 years ago, directional ITE hearing instruments have virtually disappeared from the marketplace. The cause for this may have been the manufacturing difficulties in building custom ITE instruments using the larger directional microphones that were available at that time. In the years since, smaller microphones have become available that permit their incorporation more easily in ITE instruments. Another factor that may have impacted upon the lack of success in the marketplace of early directional ITE hearing aids is the fixed amount of directionality present in the instruments that gave the wearer no frame of reference with which to judge the benefits of the directional microphone. The current study was intended to evaluate a switched dual microphone ITE hearing instrument that provided the wearer with the choice of directional or omnidirectional listening modes. Subjects METHOD Ten hearing-impaired adults, nine males and one female (mean age = 66 years, SD = 8.3 years), with bilaterally symmetric sensorineural hearing losses (air-bone gaps less than 10 db from 250 through 4000 Hz and normal tympanograms), participated in the study. The hearing losses were varied and ranged from mild to Table l Individual Right-Ear Audiograms (db HL) and Means and SDs for the 10 Subjects Frequency (Hz) Subject Mean SD

3 Directional/ ITE/Preves et al A f = 500 khz 0 Table 2 Directivity Index at 500, 1000, and 2000 Hz of Switched /Directional ITE Instruments in Directional on a 2-cm3 Coupler in an Anechoic Chamber ubject Right or Left of Binaural Set DI at 500 Hz DI at 1 khz DI at 2 khz Three- Frequency Mean B f = 1 khz f db Attenuation 9s 108 1zo R L R L R L R L R L R L R L R L R L R L Mean SD C f = 2 khz 0 Meniere's syndrome or retrocochlear disorder, nor was tinnitus a primary complaint. None of the subjects had worn hearing aids previously. Subjects were each paid $50 for participation and were allowed to keep the experimental hearing aids at the conclusion of the study. Hearing Instruments Figure l Typical polar patterns at A, 500 Hz, B, 1 khz, C, 2 khz for ITE hearing aid in directional mode on 2- cm3 coupler in an anechoic chamber. Input level is 70 db SPL. Numbers near vertical axis indicate relative gain in decibels ; numbers around circle indicate angle of incidence in degrees. moderate/severe with configurations from essentially flat to sharply sloping. Individual pure-tone audiograms for the right ears of the 10 subjects are shown in Table 1 along with the mean and standard deviation of the audiograms. All subjects were native English speakers and did not have a diagnosis of or symptoms consistent with Binaural full-shell custom ITE hearing instruments were made for each subject, each with two omnidirectional microphones operating into an electrical network to allow the wearer to convert between omnidirectional and directional operating modes via a toggle switch. Subjects were blind to the function of the switch positions, that is, they were simply aware that things might sound differently in each switch position. These devices are marketed by Micro- Tech, Inc. as the model "Persona Choice." In this design, the distance between the two microphone ports is typically 0.3" but can be varied, depending on ear size. To ensure that port axis of the directional system would be horizontal as worn, while the subject looked straight ahead, a line was drawn across the earmold impressions in the subject's ears using a protractor with a 275

4 Journal of the American Academy of Audiology/Volume 10, Number 5, May 1999 A 60 Unequalized (A) Microphone m a c C ~_ "- Directional k 2k 4k 8k Frequency in Hz Frequency (Hz) 0 deg deg deg Directional Microphone B 60 Equalized m a 20 Frequency (Hz) 0 deg deg deg Figure 3 Representative example of real-ear measurements of insertion gain at 0-, 90-, and 180-degrees azimuth with frequency responses equalized for omnidirectional (A) and directional (B) operating modes k 2k 4k Sk Frequency in Hz Figure 2 Typical frequency responses on HA-1 2-cm3 coupler in an anechoic chamber for AGC-O ITE hearing instrument differences between directional (lower curve) and omnidirectional (upper curve) modes. Frequency responses between modes (A) unequalized and (B) equalized. Input : 70 db SPL speech-shaped random noise per ANSI plumb line. The two omnidirectional microphones in each hearing aid were matched for phase and amplitude. Each ITE instrument had an automatic gain control with output compression (AGC-O) circuit with a compression threshold trimmer. Vent size was selected for each hearing aid by the Micro-Tech design program. The gain-frequency responses were also selected using Micro-Tech's design program to approximate the National Acoustics Laboratories-Revised (NAL-R) (Byrne and Dillon, 1986) formula with the volume control at one half to three-quarters rotation in the omnidirectional mode. With the toggle switch in directional mode, the instruments were designed to provide a supercardioid polar directivity pattern (Fig. 1). The supercardioid polar pattern has about 1 db greater Directivity Index (DI) than the cardioid polar pattern (Preves, 1997). The DI is the ratio of frontal energy to random energy and is approximately equivalent to the improvement in SNR that the directional instrument should provide with the wearer looking directly at the sound source. The DI for each instrument is reported in Table 2 and was determined by placing it on an HA-1 2-cm3 coupler in an anechoic chamber. For the 20 ITE hearing instruments, the mean DIs were 4.0, 5.0, and 5.2 db at 500, 1000, and 2000 Hz, respectively, resulting in an overall mean at three frequencies of 4.7 db. The standard deviations of the DIs, also reported in Table 2, were 0.8, 0.4, and 0.3 db, respectively, at 500, 1000, and 2000 Hz, indicating that the instruments were relatively consistent in the amount of directivity they provided. Unless compensated for, a directional microphone usually has less low-frequency sensitivity than the same microphone type having an omnidirectional polar pattern (Fig. 2A). This reduction in low-frequency sensitivity results from the amplitude and phase of the sound pres- 276

5 Directional/ ITE/Preves et al sure condensations and rarefactions being more similar at the two microphone ports for low-frequency waveforms than for high-frequency waveforms. In Experiment 1, the hearing aids had unequalized frequency responses for the two modes of operation, that is, the low-frequency gain was reduced in the directional mode relative to the omnidirectional mode. In Experiment 2, additional circuitry was incorporated to equalize the frequency responses, as shown in Figure 2B. Procedures Experiment 1 : Frequency Responses Not Equalized between and Directional Operating s Each subject was fitted binaurally with the hearing instruments using a Frye Fonix 6500 real-ear system to examine how well the omnidirectional microphone mode approximated the NAL-R prescriptive formula for a broadband noise input signal of 65 db SPL with the volume control between one half and three-quarters of full-on. All fittings were done by either the second or third author. If the measured hearing aid response was not a good approximation of the target (within 3 db at frequencies <_3 KHz and within 15 db at 4 khz), the hearing aids were returned to the manufacturer with the individual real-ear unaided response curves for adjustment of the frequency response and/or gain. All but one of the fittings met these criteria at the initial session. The AGC-O output limiting potentiometer was set so that the SSPL90 would not exceed the speech UCL across the speech range. The appropriateness of this setting was verified with real-ear saturation response and subjective judgment of "loud, but not uncomfortable" for a broadband input signal of 90 db SPL with the volume control at the maximum level before feedback. DIs were obtained for the directional mode of each instrument in an anechoic chamber on an HA-1 2-cm3 coupler. For comparison, real-ear aided responses in omnidirectional and directional modes were obtained for each hearing aid with a broadband noise from the Frye Fonix 6500 analyzer at 0-degrees, 90-degrees and 180- degrees azimuth, and 0-degrees elevation. A representative example of the latter measurements is shown in Figure 3. All subjects completed the Abbreviated Profile of Hearing Aid Benefit (APHAB ; Cox and Alexander, 1995) for unaided listening in the first session. The APHAB consists of 24 items that are scored in four subscales. The four subscales are EC (ease of communication in relatively quiet environments), RV (communicating in high levels of reverberation), BN (communicating in high levels of background noise), and AV (the degree of unpleasantness of environmental sounds). In the first three subscales, a higher benefit score indicates better perceived performance. Benefit is calculated as aided minus unaided score. In the fourth subscale, a less negative benefit score indicates better perceived performance. Following fitting of the hearing aids, there was an extensive individual hearing aid orientation, and subjects were instructed to wear the hearing aids 6 or more hours per day for no less than 5 days per week. They were also given blank versions of the APHAB questionnaire to take home, with instructions to complete the questionnaire for aided listening. The questionnaire forms were modified so that the wearer would judge perceptual differences between the switch in the #1 position (directional mode) and in the #2 position (omnidirectional mode) in a number of everyday quiet and noisy environments. Further, it was requested that they keep a written diary of their experiences and opinions regarding their performance with each of the switch settings. They were told that they might like one switch setting best all the time, or they might find each switch setting worked better in different listening environments, or that they might not be able to tell a difference between the switch settings at all. Because of the decrease in overall gain when the subjects switched from the omnidirectional microphone switch setting to the directional microphone switch setting in this experiment, they were informed that they might want to adjust the volume control to compensate for loudness differences when they switched between settings. Within 1 week after the fitting, subjects were either seen at the clinic or received a telephone call to determine if they had any questions about use of the hearing aids. However, no changes were made in the electroacoustics at this follow-up. After a 3- to 6-week trial period with the hearing aids, the subjects returned to the clinic and were queried about which switch setting they would prefer if they could only have one. The APHAB questionnaires for the two aided listening conditions (omnidirectional versus directional microphone mode) were retrieved as well as their written diary comments. Speech recognition measurements were made in a large 277

6 Journal of the American Academy of Audiology/Volume 10, Number 5, May 1999 Industrial Acoustics Corporation sound-treated booth, using the HINT (Nilsson et al, 1994) presented from a compact disc (CD) player. The HINT is an adaptive task in which a speech spectrum noise level is fixed and the speech level is varied in 2-dB steps to determine the SNR for approximately 50 percent sentence recognition. For this study, the HINT noise stimulus was fixed at 65 db SPL for the combined output of two loudspeakers placed at about 115- degrees azimuth and 245-degrees azimuth to avoid artificially elevating performance by placing the noise sources at the locations of the major nulls of the directional microphone polar pattern. The speech stimuli were presented from a third loudspeaker placed at 0-degrees azimuth since persons normally look in the direction of the sound source to which they are trying to listen. The two noise sources were uncorrelated in time because two different HINT CDs were used and the noise channels started at different times on the disc. The sentence lists were randomized across trials. The HINT test was given in the omnidirectional operating mode and in the directional operating mode conditions. The omnidirectional condition was first evaluated with the volume control at target gain, as validated by probe microphone measurement, and the directional condition was tested with the same volume control setting. This procedure produced a drop in overall gain for the directional condition (see Fig. 2A). Subsequently, the subjects were offered the chance to readjust the volume control settings for each microphone condition while listening to male-connected discourse in quiet at 65 db SPL if they felt that the target volume control setting was not at most comfortable listening level. If this volume control setting was changed by the subject, the HINT was also given at the new volume control settings. To increase reliability of the measurements, a second HINT list was given if there were less than three zero-axis crossings on the initial run in any condition, with the final value the mean of the two runs. Experiment 2: Frequency Responses Equalized between and Directional Operating s At the end of Experiment 1, the subject's ITE hearing aids were returned to the manufacturer for modification so that the frequency responses/gain were more closely equalized between the switch settings (see Fig. 2B). Each subject was then refitted with the instruments, probe microphone measurements were completed again by azimuth, and the subject was sent home for an additional 3- to 6-week trial period. During that period, they were again asked to fill out the modified APHAB questionnaire for each switch setting and to keep an informal diary of their experiences. The subjects were told that the loudness difference between Table 3 Individual and Mean HINT Results with Frequency Responses Not Equalized between Directional and s (Experiment 1) at Target Gain and at MCL Volume Control Setting* Volume Control at Target New Data If Volume Control Moved for MCL Subject SNR SNR Directional minus Directional SNR SNR Directional minus Directional Mean SD *Note that a smaller or more negative number indicates better performance in a given mode. Also shown are the differences in db between performances in each mode. 278

7 Directional/ ITE/Preves et al Table 4 Subjective Comments of Subjects at the End of Experiment 1 with Frequency Responses Not Equalized between Directional and s Subject Wears in directional mode usually ; reduces background noise interference sounds more "coarse" and lets in more background noise Road noise very bad in omnidirectional mode ; church service and theater better in directional mode Directional better in large, busy rooms ; uses omnidirectional only for quiet voices in small rooms Prefers directional for sales meetings and in car ; uses omnidirectional for sounds of nature Prefers directional mode since it reduces background noise while still making sound clear/good quality Likes both : directional mode is best in noise, omnidirectional mode is best in quiet Prefers omnidirectional mode in groups and watching TV, but not for loud voices or car noise Very slight preference for omnidirectional since it is louder, which makes speech clearer Prefers omnidirectional for soft voices in quiet, but directional makes speech clearer in noise the switch settings would not be so noticeable now and that their feelings about the switch settings might be the same as before or might change. At the end of the trial, the subjects returned to the clinic, the APHAB and diary notes were retrieved, and the subjects were queried regarding overall preference between switch settings. The HINTs, as described above, were repeated. In addition, because equalization allowed quick comparisons without volume control changes, paired-comparison testing of perceptual judgments was completed in the soundbooth. The experimenter sat behind the patient and switched back and forth between directional and omnidirectional modes on the hearing aids with approximately a 5-second listening time at each setting, querying the subject about which sounded best for several criteria. While listening to the male-talker, continuous discourse in quiet at 65 db SPL, the subjects were queried regarding the "clarity or clearness of the speech" and the "sound quality, pleasantness, or naturalness of the sound." When listening to maletalker continuous speech at 65 db SPL, presented in multitalker babble background noise from the loudspeakers at 115- and 245- degrees azimuths at 57 db SPL (+8 db SNR), the subjects judged the criteria above, but also were queried regarding "how bothersome, annoying, or interfering the background noise is." Subjects were asked whether switch setting 1 or 2 was better for a given criterion, or if there was no difference between them. The order of criteria, listening condition (quiet, noise), and switch setting (1-2 versus 2-1) was randomized. the end of each listening condition, a final criterion was added : the subject was asked which switch setting was better "overall, considering all the criteria and what is most important to you." Five paired comparisons were done in each condition for each criterion. RESULTS Experiment 1: Frequency Responses Not Equalized between and Directional Operating s Speech Recognition in Noise At en GD 20 b -20 EC RV BN AV Subscale "omnidirectional ID Directional Bars are 1 std. d- Figure 4 Group mean APHAB benefit scores for each subscale and operating mode from Experiment 1 (with frequency responses not equalized between directional and omnidirectional modes). Table 3 shows the individual and mean HINT results (in db SNR) for omnidirectional mode and directional mode for the target volume control settings. Note that a smaller or more negative number indicates better performance. Mean SNRs in omnidirectional mode and directional mode were -1.2 db and -4.0 db, respectively. This indicates that, on average, 2.8 db less speech level was required in the directional mode as compared to the omnidirectional mode in the given noise level in order to achieve 50 percent correct sentence recognition. This difference was statistically significant at the p =.0007 level on a paired t-test. 279

8 Journal of the American Academy of Audiology/Volume 10, Number 5, May 1999 Table 5 Individual and Mean HINT Results with Frequency Responses Equalized between Directional and s (Experiment 2) at Target Gain and at MCL Volume Control Setting* Volume Control at Target New Data If Volume Control Moved for MCL Subject SNR SNR Directional minus Directional SNR SNR Directional minus Directional Mean SD *Note that a smaller or more negative number indicates better performance in a given mode. Also shown are the differences in db between performances in each mode. When offered the opportunity to change the volume control settings in each microphone mode for greater listening comfort, 5 and 6 of the 10 subjects did not change the settings in omnidirectional and directional microphone modes, respectively, indicating that it was already comfortable as set. For those five subjects who changed the setting in omnidirectional mode, all chose to decrease the gain, albeit often slightly, from the target setting. For the directional mode, two subjects chose to increase gain slightly and two chose to decrease gain slightly. Shown in the second part of Table 3 are the HINT results for the volume control settings at these new (MCL) values. Note that in no case was the relative performance ranking of omnidirectional versus directional changed, that is, if directional performance was better at target volume control setting, it was also better at MCL. Questionnaire Results Figure 4 shows the mean and standard deviation APHAB benefit scores for the omnidirectional and directional modes. These data are consistent with those reported by Cox and Alexander (1995) for linear hearing aid users. Paired t-tests were performed on the data under each of the four subscales of the APHAB. For the EC subscale, subjects showed a 2.4 percent mean improvement in perceived performance in the directional mode as compared to the omnidirectional mode, but this difference was not statistically significant. For the RV subscale, subjects showed a 4.5 percent mean improvement in the directional mode as compared to the omnidirectional mode. This difference was statistically significant at the p <.05 level. For the BN subscale, subjects showed a nonsignificant EC RV BN AV Subscale 10 s JS 6 ~ _ 7 NOmnidimctions m Directional N 6 Bars are 1 std. dev. c 5 ig 4 E 3 i "Clan ' a xs~d Noise" Quiet Noise Quiet Noise Quiet Noise Quiet Noise Condition Omnidiredional "Directional Figure 5 Group mean APHAB benefit scores for each subscale and operating mode from Experiment 2 (with frequency responses equalized between directional and omnidirectional modes). Figure 6 Number of subjects (of 10) who indicated a significant preference (three or more choices of five) for directional or omnidirectional modes in paired-comparison testing. 280

9 Directional/ ITE/Preves et al Table 6 Subjective Comments of Subjects at the End of Experiment 2 with Frequency Responses Equalized between Directional and s Subject 9 10 Believes always hears better in directional mode and leaves it set there all of the time Prefers directional for TV with kids around and usually leaves it set there Prefers directional for hearing better at church, for TV, and usually leaves it set there Prefers directional but cannot explain why No preference ; no difference detected between settings this time Prefers directional ; it picks up less environmental noise ; omnidirectional is too loud sometimes No preference ; likes both ; omnidirectional is clearer in quiet, directional is clearer in noise Prefers omnidirectional ; is clearer for speech, but directional reduces background noise better in some situations Directional is more natural sounding, especially in noise, but omnidirectional is better for wife's soft voice No preference ; directional makes speech clearer in crowds but omnidirectional better in one-on-one conversation 3.6 percent mean improvement in the directional mode as compared to the omnidirectional mode. For the AV subscale, subjects showed a nonsignificant mean improvement of 1.7 percent in the omnidirectional mode as compared to the directional mode. Subjective Comments Table 4 lists some of the major comments made by the 10 subjects comparing the directional mode to the nondirectional mode of operation with the frequency responses between omnidirectional and directional operating modes not equalized. If forced to select only one microphone mode, 6 of the 10 subjects said they would prefer the directional mode because it reduced background noise, while 3 subjects said they would prefer the omnidirectional mode. The other subject had no preference, perceiving little difference between the directional and omnidirectional operating modes. Most of the subjects, however, reported using both operating modes, depending on the listening situation, and said they would be hesitant to give up either one. Experiment 2: Frequency Responses Equalized between and Directional Operating s Speech Recognition in Noise Table 5 shows the individual and mean HINT results (in db SNR) for the omnidirectional and directional modes for the target volume control settings with the frequency responses equalized between directional and omnidirec- tional operating modes. Mean SNRs in the omnidirectional and directional modes were -1.9 db and -4.3 db, respectively. This indicates that, on average, 2.4 db less speech level was required in the directional mode as compared to the omnidirectional mode in the given noise level in order to achieve 50 percent correct sentence recognition. In a paired t-test, this difference was statistically significant at the p =.01 level. When offered the opportunity to change the volume control settings in each microphone mode for greater listening comfort, 6 of the 10 subjects did not change the settings in either omnidirectional or directional microphone modes, indicating that it was already comfortable as set. For those four subjects who changed the setting in omnidirectional mode, all chose to decrease the gain, albeit often slightly, from the target setting. For the directional mode, three subjects chose to decrease gain slightly and only one chose to increase gain. Shown in the second part of Table 5 are the HINT results for the volume control settings at these new MCL values. Note that in no case was the relative performance ranking of omnidirectional versus directional changed except for subject 10, who chose to increase gain slightly in the directional mode. For this subject, HINT performance was better in the omnidirectional mode at target volume control settings, but with the increase in gain to MCL in the directional mode, performance now exceeded that of the omnidirectional mode. Questionnaire Results Figure 5 shows the mean APHAB benefit scores for the omnidirectional and directional

10 Journal of the American Academy of Audiology/Volume 10, Number 5, May 1999 modes. Paired t-tests were performed on the data under each of the four subscales of the APHAB. For the EC subscale, subjects showed a 6.1 percent mean improvement in perceived performance in the directional mode as compared to the omnidirectional mode, but this difference was not statistically significant. For the RV subscale, subjects showed an 8.4 percent mean improvement in the directional mode as compared to the omnidirectional mode. For the BN subscale, subjects showed an 8.5 percent mean improvement in the directional mode as compared to the omnidirectional mode. The differences for the RV and BN subscales were statistically significant at the p <.05 level. For the AV subscale, subjects demonstrated a nonsignificant mean improvement of 1.9 percent in the directional mode as compared to the omnidirectional mode. Subjective Comments Table 6 lists some of the major comments made by the 10 subjects from their observations of the comparisons between the directional and nondirectional modes of operation with the frequency responses equalized. If forced to select only one mode, 6 of the 10 subjects said they would prefer the directional mode for reducing background noise, while one subject preferred the omnidirectional mode. Three other subjects had no strong preference, perceiving little difference between the directional and omnidirectional operating modes. Most of the subjects reportedly used both operating modes, depending on the listening situation, and, again, said they would prefer to have the choice. Paired Comparisons Figure 6 shows the results of the paired-comparison tests for clarity, quality, and overall performance in quiet and in noisy listening conditions and for background noise annoyance in the noisy listening condition, across the two operating modes. Shown are the number of subjects who selected the directional or omnidirectional mode in three or more of the five comparisons for a given listening condition and criterion. For that reason, the number of subjects does not always add up to 10, with the remaining subjects selecting neither mode a significant number of times and/or reporting no perceived difference. These data indicate that most of the subjects preferred the directional mode for clarity, quality, and less annoyance from background noise in noisy listening situations. The exception is clarity in quiet. DISCUSSION AND CONCLUSION T he HINT results agree with the majority of the subjective preferences for the directional mode for both the equalized and unequalized experiments. The mean reduction in SNR required for 50 percent correct sentence recognition in the directional mode relative to the omnidirectional mode was slightly greater when the frequency responses were not equalized (2.8 db) between the two modes than when they were equalized (2.4 db). One could predict that this mean SNR improvement of about 2.5 db could produce a multiplier constant of approximately 2.5. The improvement in speech recognition scores can be estimated by the following function : I = kx where I is the percent improvement, k is the multiplier constant, and x is the slope in percent/db of the Performance-SNR function (e.g., Killion et al, 1998). For the HINT materials, the slope is 8.5 percent/db for normalhearing persons with the noise presented at 90- degree and 270-degree azimuths. Using this figure as a first estimate with these data from hearing-impaired persons, the predicted mean improvement in speech recognition scores in the directional mode is approximately 2.5*8.5 or 21 percent. Because of the drop in overall level in directional mode produced by the reduced low- and mid-frequency response (see Fig. 2A) for Experiment 1, the improvement in SNR for directional mode is particularly interesting, considering that the volume control was set to target and was not changed between omnidirectional and directional modes. Some might argue that this improvement was due to reduced upward spread of masking, but the re-addition of the lower frequency energy in Experiment 2 did not change the results substantially. The difference in HINT scores between directional and omnidirectional modes summarized in Table 3 for the unequalized responses can be compared to the data from Valente et al (1995), since the BTE instrument used in that study also had unequalized responses between operating modes. When the hearing aids were set in the omnidirectional mode, the mean HINT SNR for 50 percent correct in this study was -1.2 db compared to a 0-dB mean in the Valente et al study. When the hearing aids were set in the directional mode, the mean HINT SNR for 50 percent

11 Directional/ ITE/Preves et al correct in this study was -4 db, which was poorer than the -7- to -8-dB means in the Valente et al (1995) study. The larger benefit provided by the directional mode in the Valente et al study may be partially explained by differences in the noise loudspeaker locations used in the two studies. In the Valente et al study, a single noise loudspeaker was placed at a 180-degrees azimuth, a location in the null provided by the polar pattern of the directional BTE hearing aid used. In this study, two loudspeakers producing uncorrelated noise were placed at 115- and 245-degree azimuths, locations that were not directly in the nulls produced by the supercardioid directivity pattern. Had the Valente et al study used multiple noise sources at locations away from the nulls in the polar directivity pattern, less benefit may have been observed in the directional mode. The APHAB results in this study also agree with the subjective comments for both the equalized and unequalized experimental conditions. Although some of the APHAB results that indicated a trend favoring the directional mode to the omnidirectional mode did not reach the level of statistical significance, most of the subjects preferred the performance of their ITE instruments in the directional switch mode position overall. One of the subjects (10) who preferred the omnidirectional mode in the unequalized frequency response condition had no preference for either mode when the frequency responses were equalized. The subjects' reports that they use both the omnidirectional and directional operation modes indicate that wearer control of the directional mode is a critical feature contributing to overall hearing aid benefit. This result is in agreement with that of Kuk (1996). Hearing aid wearers encounter varied listening environments and, for most of the subjects in this study, neither the omnidirectional or the directional operation mode alone was preferred over having access to both modes. There may be certain groups of hearing aid wearers for whom access to only a single operational mode may actually create a significant detriment in speech perception and hearing aid benefit. For example, by providing only the directional operation mode in hearing aids fit to a school-aged child, the child may be at a distinct disadvantage for listening to speech (e.g., from other children in a classroom) presented from sources located to the sides and/or behind the child. As there is little or no information regarding the speech perception performance of dual-microphone directional hearing aids when speech is presented from sources located between 90-degrees and 180- degrees azimuth, access to a wearer-actuated switch controlling operational modes becomes paramount. This is particularly true if there are unequalized frequency responses between the omnidirectional and directional modes. In conclusion, ITE hearing instruments can be made with directional performance with dual microphones that is comparable to directional BTE instruments. A dual-microphone ITE instrument incorporating switch-selectable directional/omnidirectional operating modes may provide more perceived benefit than the ITE directional instruments of 20 years ago by eliminating the problems of having a directional response in all listening situations and not having a frame of reference with which to gauge benefit. Acknowledgment. The authors thank Micro-Tech, Inc. for supplying the hearing aids used in this study and Mark Bren for building them ; Julie Patterson who assisted in subject scheduling, calibration checks, and data collection ; and Mary MacRae and Timothy Peterson for their assistance in preparing the figures. Results of this study were presented as a poster session at the September 1997 NIDCD Hearing Aid Research and Development Conference, Bethesda, MD. REFERENCES Byrne D, Dillon H. (1986). The National Acoustic Laboratories' (NAL) new procedure for selecting the gain and frequency response of a hearing aid. Ear Hear 7: Chasin M. (1994). Improving signal-to-noise ratio with directional microphones. Hear Instrum 45(9) : Cole W. (1977). HearingAid Having Adjustable Directivity. U.S. Patent 4,051,330. Cox R, Alexander G. (1995). The Abbreviated Profile of Hearing Aid Benefit. Ear Hear 16 : Gage F, Mutlick R. (1974). Hearing Aid With Valued Dual Ports. U.S. Patent 3,798,390. Johanson D, Stutz R. (1974). Hearing Aid Hauing Selectable Directional Characteristics. U.S. Patent 3,836,732. Killion M, Schulein R, Christensen L, Fabry D, Revit L, Niquette P, Chung K. (1998). Real-world performance of an ITE directional microphone. Hear J 51(4): Knowles H, Carlson E. (1973). HearingAid System. U.S. Patent 3,770,991. Kuk F. (1996). Subjective preferences for microphone types in daily listening environments. Hear J 49(4) : Lentz W (1972). Speech discrimination in the presence of background noise using a hearing aid with a directionally-sensitive microphone. Maico Audiol Lib Ser 10(9) :

12 Journal of the American Academy of Audiology/Volume 10, Number 5, May 1999 Mueller G, Grimes A, Erdman S. (1983). Subjective ratings of directional amplification. Hear Instrum 34(2) :14. Nielsen H. (1973). A comparison between hearing aids with directional microphone and hearing aids with conventional microphone. Scand Audiol 2: Nilsson M, Soli S, Sullivan J. (1994). Development of the Hearing In Noise Test for the measurement of speech reception thresholds in quiet and in noise. JAcoust Soc Am 95: Preves D. (1976). Directivity of in-the-ear aids with nondirectional and directional microphones. Hear J 29(8):7, Preves D. (1997). Directional microphone use in ITE hearing instruments. Hear Rev 4(7) : Rumoshovsky J. (1977). Directional microphones in ITE aids. Hear J 30(6):11. Valente M, Fabry D, Potts L. (1995). Recognition of speech in noise with hearing aids using dual microphones. JAm Acad Audiol 6: