Tech Topic | January 2017 Hearing Review
Results from an experiment using a narrow-directionality (binaural beamformer) feature indicate the advantage of this type of directional feature in situations where listening to a frontal target speech is a challenging task, and where a closed fitting is acceptable for the hearing aid wearer.
It is well established that people with sensorineural hearing loss have more difficulty than normal hearers in understanding speech in many noisy places.1 It has also been shown that, compared to normal hearers, hearing-impaired listeners require significantly more mental effort to attend to, and understand, an auditory message.2 Although hearing aids significantly reduce the mental demands in listening,3 communicating in noise remains considerably challenging, particularly in poor signal-to-noise ratio (SNR) listening conditions.
Directional features in hearing aids have been shown to significantly reduce the effort exerted when listening in noise.1,3,4-8 Although directional features can degrade sound localization, listeners appear to tolerate this in order to obtain the improved sentence recognition in noise and lower listening effort these features provide.9 Directional features in hearing aids have been shown to provide greater acceptance of hearing aids in everyday listening compared to omnidirectional microphones.7 Furthermore, it has been postulated that individuals with a greater degree of hearing loss, who are consequently least able to use localization cues, and who have greater difficulty listening in noise, are more likely to benefit from highly directional amplification strategies.6
The focus of this experiment was to examine the effect of a specific narrow-directionality feature (also known as a “binaural beamformer” or “super-directional microphone”) on speech intelligibility and self-rated listening effort, tested in extremely challenging listening conditions.
Participants. A total of 9 hearing-impaired participants with an average age of 68 years (range 55 to 78 years) and average 4FAHL500,1000,2000,4000Hz of 57 dB HL (range 38 to 84 dB HL) participated. The participants’ impairment was non-congenital in nature and symmetrical between ears. All were hearing aid wearers.
Another 9 participants who did not wear hearing aids, approximately matched in age, were recruited as “normal hearing” controls. Their average age was 62 years (range 42 to 78 years). Their 4FAHL was 10 dB HL (range 6 to 14 dB HL). Hearing threshold levels (HTLs) between 500 and 2000 Hz inclusive were all ?25 dB HL; however, some individuals exhibited mild hearing loss at higher frequencies. Given the age range of participants, this was typically age-appropriate. The pure-tone HTLs of both groups are shown in Figure 1.
Hearing aids. Hearing-impaired participants were binaurally fitted with receiver-in-the-canal (RIC) behind-the-ear instruments, model Pure 7px M. Hearing aids were coupled with fully occluding “double” domes. The hearing instruments were programmed for the fitting formula “primax™ fit,” “experienced” level. Only minor overall gain adjustment were made, according to participants’ listening comfort.
Two device microphone conditions were tested: omnidirectionality and narrow directionality. The narrow-directionality feature is commercially available in Sivantos primax™ hearing aids, and is promoted as providing a higher degree of directivity than conventional directional microphones. The narrow directionality feature uses bi-directional wireless transmission of audio between binaural hearing aids to produce a binaural beamformer output in each ear.5 This feature was manually switched on and off during testing. Switching between the omnidirectional and narrow-directional settings was made using an App, running on a smart phone. The smart phone was connected to the hearing instruments via a product-specific wireless interface system (EasyTek). Normal-hearing participants wore no hearing aids during testing.
Stimuli. The target speech was recorded in BKB-like sentences spoken by one Australian male talker and presented from a frontal loudspeaker (0° azimuth). The noise was also constructed from BKB-like sentences, spoken by the same talker as the target speech, but presented uncorrelated from seven surrounding loudspeakers at 45°, 90°, 135°, 180°, 225°, 270° and 315°. All loudspeakers were 1.2m from the listener at approximate ear level in a room for which the critical distance for speech-weighted noise was on average 0.6 meters. In addition, cafeteria background noise was added to the noise signal at 15 dB below the BKB-like noise level. The similarity of the target and noise speech was designed to make the listening task challenging for both normal-hearing and hearing-impaired participants.
Two SNR conditions were tested: 1) An individually determined SNR, which achieved a sentence recognition score of 80% in the omnidirectional condition (hearing-impaired participants), and 2) A fixed -3.5 dB SNR (all participants).
In order to determine the individual SNR for each participant with hearing impairment, the noise was presented at 66.5 dB SPL, and the level of the target speech was varied adaptively to determine the SNR for the 80% correct point, using morphemic scoring (ie, complete words or meaningful parts of words) applied to all words in the sentence. Participants were then presented with 55 sentences at this individually determined SNR, again using morphemic scoring. Once the task was completed, participants were asked to rate their listening effort during the test run, using a 13-point scale, with seven labels ranging from “no effort” to “extreme effort” (with a center rating of “moderate effort”). This task was completed by the hearing-impaired group for both omnidirectionality and narrow-directionality features, with the order of testing counterbalanced across participants.
For the fixed -3.5 dB SNR test, the noise was presented at a fixed level of 66.5 dB SPL, and the target speech at 63 dB SPL. Participants were again presented with 55 sentences, and rated the listening effort at the end of the test run. This testing was completed by the hearing-impaired group with the narrow directionality feature, and by normal hearers without hearing aids.
The order of test conditions was counterbalanced across participants.
Administration of the adaptive test to determine-the individual SNR for 80% correct (using the omnidirectional microphone) resulted in SNRs ranging from -3 dB to +9 dB.
Figure 2 shows the percent-correct scores for participants with hearing impairment using the omnidirectional and the narrow-directionality features. A paired t-test on the arcsined scores indicated that the narrow-directionality scores were significantly higher than the omnidirectional scores (p < 0.002). A ceiling effect for the narrow directionality condition is evident. According to test scores, it also appears that the initial adaptive procedure overestimated the performance of one participant (who was tested at -3 dB SNR). Even for this participant, however, the score in the narrow directionality condition was close to ceiling.
Figure 3 shows the ratings of listening effort given by each participant for each test condition. On average, the participants rated listening effort 4 scale points lower for the narrow-directionality than for the omnidirectional microphone feature. For omnidirectionality, the average rating was considerable effort, whereas for narrow-directionality it was little effort. A paired t-test indicated that the difference in the ratings was significant (p=0.001).
Figure 4 shows the scores when both the hearing-impaired and normal-hearing participants were tested at -3.5 dB SNR. The figure makes it evident that the performance of the hearing-impaired group matched that of the normal hearers for all but one participant with severe hearing loss. The regression line fitted to the results for the hearing-impaired group accounts for 64% of the variance, with r = -0.83, p = 0.006. When listening with narrow-directionality, for every 10 dB increase in four frequency average (4FA) hearing loss, the sentence recognition scores decreased by 9.3%. Given that the slope of the psychometric function is about 9% per dB (based on data obtained on normal-hearing participants but not reported in this article), each additional 10 dB of hearing loss would, on average, result in a need for an additional 1 dB improvement in SNR to remain at the same speech reception threshold performance levels.
Figure 5 shows that rated listening effort was similar for both groups of listeners, both in terms of mean scores and the range of scores. A paired t-test indicated that the difference in the rated effort was not significant (p = 0.9).
Although the slope of the regression between 4FAHL and listening effort indicated greater listening effort for greater degree of hearing loss, this was not statistically significant (r = 0.64, p = 0.066).
Summary and Implications
The implications of these results for real-world listening situations are compelling; however, it is acknowledged that this experiment evaluated listening to frontal target speech surrounded by a contrived speech babble from other horizontal directions. Nevertheless, most listeners with moderate hearing loss, when aided with narrow-directionality, understood speech as well as approximately age-matched (near-) normal- hearing listeners without hearing aids. Based on the regression line in Figure 4, a hearing-impaired listener with a 4FAHL of 52 dB HL would, on average, hear as well using this technology in this listening situation as the average age-matched person without marked hearing loss. While the regression line clearly cannot extend above 100% word recognition, it does seem likely that listeners with mild hearing loss using the same technology (and a fully occluding ear fitting) may outperform (near-) normal-hearing listeners in difficult listening situations. This is currently under further evaluation.
Similarly, based on the regression line in Figure 5, a hearing-impaired listener with a 4FAHL of 57 dB HL would, on average, exert the same listening effort as an average age-matched (near-) normal hearer in this same listening situation. These results are consistent with earlier reports10 of the significantly greater benefit gained from narrow directionality compared to omnidirectional microphones, both in term of increased intelligibility and reduced listening effort.
It is acknowledged, however, that the benefits of narrow-directionality measured in this experiment were obtained with double-dome fittings, which typically occlude the ears. Such occluded ears often cause an unacceptable own-voice quality to people with low frequency hearing thresholds, better than approximately 45 dB HL. If the narrow-directionality feature is applied with an open fitting, there will be no directivity benefit in the frequency region for which sound directly entering the ear canal dominates the electronically amplified sound. As with conventional directional microphones, the benefit of directivity would therefore reduce as the bandwidth of the direct-sound region increases.
In summary, the technology evaluated demonstrated a significant listening advantage in a laboratory experiment comprising a frontal target speech in diffuse speech babble. For this experiment, hearing-impaired participants with moderate hearing loss matched the intelligibility performance of a group of age-matched near-normal hearers. The range of effort incurred in listening for the hearing-impaired participants also matched that of the age-matched, near-normal hearing group.
The results of this experiment demonstrate the advantage of this type of directional feature, in situations where listening to a frontal target speech is a challenging task and where a closed fitting is acceptable for the hearing aid wearer.
This research was performed by the authors under contract with Sivantos. All measurements, analysis, and reporting were performed by National Acoustic Laboratories (NAL) staff. We would like to thank Els Walravens, Mark Seeto, Tim Beechey, Margot McLelland, and Fabrice Bardy for performing some of the measurements.
Kramer S, Kapteyn T, Houtgast T. Occupational performance: Comparing normally-hearing and hearing-impaired employees using the Amsterdam Checklist for Hearing and Work. Int J Audiol. 2006;45:503-512.
Desjardins JL, Doherty KA. Age-related changes in listening effort for various types of masker noises. Ear Hear. 2013;34(3):261-272.
Hornsby BWY. The effects of hearing aid use on listening effort and mental fatigue associated with sustained speech processing demands. Ear Hear. 2013;34(5):523-534.
Desjardins JL. Analysis of performance on cognitive test measures before, during, and after 6 months of hearing aid use: A single-subject experimental design. Am J Audiol. 2016;25(2):127-41.
Kamkar-Parsi H, Fischer E, Aubreville M. New binaural strategies for enhanced hearing. Hearing Review. 2014;21(10):42-45.
Kidd Jr G, Mason CR, Best V, Swaminathan J. Benefits of acoustic beamforming for solving the cocktail party problem. Trends in Hearing. 2015;19:1-15.
Picou EM, Aspell E, Ricketts TA. Potential benefits and limitations of three types of directional processing in hearing aids. Ear & Hear. 2014;35:339-352.
Quintino CA, Mondelli MFCG, Ferrari DV. Directivity and noise reduction in hearing aids: speech perception and benefit. Braz J Otorhinolaryngol. 2010; 76(5):630-638.
Best V, Mejia J, Freeston K, Van-Hoesel R, Dillon H. An evaluation of the performance of two binaural beamformers in complex and dynamic multitalker environments. Int J Audiol. 2015;54(10):727-735.
Froehlich, M, Freels K, Powers TA. Speech recognition benefit obtained from binaural beamforming hearing aids: Comparison to omnidirectional and individuals with normal hearing. May 28, 2015. Available at: http://www.audiologyonline.com/articles/speech-recognition-benefit-obtained-from-14338
Correspondence can be addressed to HR or Dr Mejia at: [email protected]
Original citation for this article: Mejia J, Carter L, Dillon H, Littman V. Listening Effort, Speech Intelligibility, and Narrow Directionality. Hearing Review. 2017;24(1):22.