By Matthias Latzel, PhD, and Jennifer Appleton, MSc
As the number of hearing aid wearers who lead an active lifestyle increases, many of these consumers are now facing the difficulty of listening in windy environments. Phonak’s Speech in Wind feature replaces the low-frequency areas that are disrupted by wind on one side of the head with useful low-frequency information from the other side where less wind is present. This paper looks at a new hearing aid design to improve speech intelligibility in windy situations.
Wind Noise Strategies
Wind noise is often a problem for hearing aid wearers—particularly for those people who enjoy a lot of outdoor activities or for those who work outdoors. Wind noise can be produced either passively, as a result of windy conditions, or actively, by the actual hearing aid being moved (eg, jogging, bike riding, etc).
Turbulence created by wind is picked up by the hearing aid microphone, generating wind noise at the hearing aid output. Lower wind speeds generally produce wind noise within the lower frequency range. In comparison, higher wind speeds tend to produce wind noise energy higher in frequency and also produce higher levels of noise in general.1-3
Reports indicate that only about 50% of hearing aid users are satisfied with their hearing aid’s performance in situations where wind noise is present.4 Given that the number of hearing aid users who lead an active lifestyle is increasing, there is a need to improve both the comfort and speech intelligibility within windy situations.
Physical modifications. There have been several attempts to reduce wind noise in hearing aids. Physical modifications to the hearing aids, such as adding a hood on top of the microphone or covering the microphone with a thin layer of foam, have been implemented although the effectiveness of such measures is unknown.5 Other physical modifications have involved an alteration of form factor, such as an in-the-ear hearing aid with a microphone placed in the small indentation between the crura of helix and antihelix in the pinna. The aim of this was to shield the microphone from wind. Again, the success of such methods has not been verified in peer-reviewed literature.
Wind noise algorithms. A range of algorithms have been implemented in hearing aids in order to attempt to differentiate between wind noise and a desired signal. These algorithms typically involve a reduction in amplification at certain frequencies. Although this has improved comfort in listening in wind, it has not generally improved speech intelligibility because part of the useful speech signal is also reduced.
New Solutions for Wind Noise
In the Phonak Spice+ platform, the wind noise canceller (WNC), WindBlock, can detect wind noise and apply a reduction of gain below 1000 Hz. This improves comfort for the hearing aid wearer in windy situations. In Phonak’s Quest platform, the enhanced WNC is more advanced. It now uses information from both microphone ports on a hearing instrument in order to detect wind noise. In addition to this, the low frequency attenuation in the Quest platform is up to 6.5 dB higher than that within the Spice+ platform.
Enhanced WNC in Quest has a higher dynamic range for wind speed detection. The relationship between attenuation of wind noise and wind speed is much better than in the Spice+ platform products. In other words, enhanced WNC detects wind faster and the resolution in which it cancels wind is also higher. This can be seen in Figure 1.
|
| Figure 1. The enhanced WNC in Quest has better resolution of different wind speeds and more low-frequency attenuation than the WindBlock within the Spice+ platform. (click to enlarge) |
Phonak’s Speech in Wind (SiW) feature within the Quest platform is designed to improve not only comfort, but also speech intelligibility in windy situations. This is a binaural signal processing algorithm based on the principle of binaural hearing. An example of a binaural hearing strategy would be the ‘’better hearing’’ effect. The human auditory system uses the signal from the ear that has the better SNR while attenuating the signal from the other ear with worse SNR.
The SiW feature mimics this concept. When Phonak Quest hearing aids are worn in both ears, the low frequency part of the microphone signal from the ear where more wind noise is present is substituted with the low frequency part of the microphone signal from the other ear where less wind noise is present. The high-frequency component of the signal is unaffected.
In order to mimic the subconscious nature of natural binaural hearing, the SiW feature is automatically activated whenever the system detects an asymmetric wind situation. The user does not have to press any buttons in order to activate it.
When wind noise of 0 to 4.5 meters per second (m/s, or about 0 to 10 miles per hour) is detected in one hearing instrument, it checks with the other hearing instrument to see how much wind noise is present. If wind is detected in both hearing instruments but there is a difference in this wind level, then the microphone signal of the hearing aid most affected by the wind noise is attenuated. This hearing aid then receives the complete streamed signal from the hearing aid with less wind noise.
This streaming process is demonstrated in Figure 2. Only the low frequency (below 1.5 kHz) part of this streamed signal is combined with the high frequency part of the original signal as the high frequencies are less affected by wind noise.
|
| Figure 2. The audio signal from the hearing instrument with the least wind noise is streamed to the hearing instrument with the most wind noise. |
This provides a better signal-to-noise ratio (SNR) in order to understand speech. As the higher frequency sounds are unaffected, spatial cues needed for localization are preserved.
Clinical Verification
Verification of the Speech in Wind feature was carried out at Phonak AG in Stäfa, Switzerland. For the SiW feature, only the frequency range below 1.5 kHz should be incorporated from the hearing aid with the better SNR into the hearing aid with the worse SNR in order to improve speech intelligibility. This means that binaural differences in level and phase should be preserved in the high frequency range.
|
|
| Figure 3. Comparison of the polar diagrams of the ipsilateral (right – better SNR) and the contralateral (left – poorer SNR) hearing aids. Top: Polar diagrams for the 160 Hz, 320 Hz, 480 Hz, 640 Hz, and 800 Hz frequencies. The polar patterns are identical so the signal is streamed from right to left. Bottom: Polar diagrams for the 2,160 Hz, 2,480 Hz, 2,880 Hz, 3,360 Hz, and 3,920 Hz frequencies. The polar patterns are different so at these frequencies the original signal is used. And no streaming is going on. (click to enlarge) |
In order to verify this, a Knowles Electronic Manikin for Acoustic Research (KEMAR) was set up wearing binaural Phonak Quest behind-the-ear hearing aids. Figure 3 shows the polar diagrams measured on the KEMAR in low (top diagrams) and high frequencies (bottom diagrams) for both hearing aids respectively. The polar diagrams for the low frequencies are clearly identical, which confirms that the signal components are transferred from the hearing aid with the better SNR (right-ipsilateral) to the hearing aid with the poorer SNR (left-contralateral). On the other hand, the bottom diagrams demonstrate that no overlaying of signal has occurred for high frequencies, since the right and left polar patterns look clearly different from each other.
Further verification of the Speech in Wind feature was carried out by King Chung, PhD, at the University of Illinois.5 One of the purposes of the study was to compare the effectiveness of wind noise algorithms implemented in digital hearing aids with standard WNC and the new Speech in Wind approach.
Two digital behind-the-ear (BTE) hearing aids were programmed to have matching frequency responses when worn on a KEMAR. The hearing aids were set up as follows: Spice+ with WNC activated and deactivated, and Quest with Speech in Wind activated or deactivated. The different settings within the hearing aids were accessible by changing to different listening programs using remote controls. The frequency responses of all hearing aids were programed to NAL-NL2. A standard audiogram was used (E DIN EN 60). The hearing aids were programed to omnidirectional modes in all experimental conditions because most currently available digital hearing aids switch to the omnidirectional mode when wind is detected.
All testing was carried out in a quiet wind tunnel built for acoustic research.2,6 A fan produced wind velocities in the tunnel of 0 m/s, 2.5 m/s (approximate speed of a brisk walk), 4.5 m/s (approximately the average wind velocity in the United States on non-windy days), 9.0 m/s, and 13.5 m/s (wind velocity for the National Weather Service to issue strong wind advisory).
The hearing aids were worn on the KEMAR, which was turned gradually in 10° steps in the wind tunnel. Hearing aid outputs were recorded at each of these 10° increments from 0° to 360°.
Overall levels and one-third octave band levels of the wind noise recordings were analyzed. Figure 4 shows the difference in Sound Pressure Level (SPL) when the WNC is on versus when it is off, at a wind speed of 4.5 m/s (average wind velocity on non-windy days). The flat, green area indicates no difference in output signal. If the peak is yellow or red, this indicates that the output without the WNC is higher and therefore the WNC is effective. As there is a greater area of yellow and red peaks for the Quest device, this indicates that the new approach has a more effective WNC than that of the Spice+ device.
|
| Figure 4. For the wind speed of 4.5 m/s, each graph shows the difference in Sound Pressure Level (SPL) when the WNC is on versus when it is off. A flat, green area indicates no difference in output signal. If the peak is yellow or red, this indicates that the output without the WNC is higher and therefore the WNC is effective. (click to enlarge) |
|
| Figure 5. The difference in SPL when the Quest binaural part of the WNC is compared to just its monaural part. A flat, green area indicates no difference in output signal. If the peak is yellow or red, this indicates that the output is higher when the binaural part of Speech in Wind is active. (click to enlarge) |
After establishing that the newer WNC is more effective than the older WNC, it was then interesting to investigate whether the binaural part of Speech in Wind provides extra benefit. Figure 5 shows the difference in SPL output when the binaural part of Speech in Wind is (automatically) activated. A flat, green area indicates no difference in output signal. If the peak is yellow or red, this indicates that the output is higher when the binaural part of Speech in Wind is active.
Given that there are some yellow peaks in Figure 5, this indicates that the output of the binaural part of Speech in Wind is higher at low frequencies at certain angles. At first glance, this seems to suggest that Speech in Wind is actually less effective in this condition, as it could be assumed that more wind noise is present. However, this is not the case; when using the binaural part of Speech in Wind, the low frequency signal from the hearing aid with less wind noise is streamed to the hearing aid with more wind noise. This “useful” low frequency signal (containing also speech cues if speech is present) replaces the “less useful” ( ie, more masked by wind) signal. Therefore, this results in a higher output in comparison to just using the monaural part of Speech in Wind where the low frequency signal is simply attenuated. Less low-frequency energy is not necessarily better, because not only is the wind noise attenuated, but so are low-frequency components of speech, which results in speech sounding less full.
In Figure 5, green flat areas indicate that the binaural part of Speech in Wind does not provide any changes. This is intentional because at these angles no clearly asymmetric differences in wind noise are occurring and therefore in these cases, the binaural part is automatically switched off and only the monaural part is active. The binaural part is actually only active at particular angles because of the special wind flow conditions of the bald KEMAR. In a real test subject with hair and soft surface (skin), the flow of wind between the ears is lower in comparison to a bald head which has a hard surface, such as the KEMAR. Therefore, in real subjects, there tend to be more angles where there is an asymmetry in wind noise between ears and therefore the binaural part of Speech in Wind is activated more often in real subjects.
Conclusion
These verification studies confirm that the Phonak Speech in Wind feature (binaural WNC) works as intended. Both studies indicate that low frequency signal components are transferred from the hearing aid with the better SNR (less wind noise) to the hearing aid with the poorer SNR (more wind noise). As high frequency signal components are not affected, localization ability is maintained.
In order to confirm the benefit of this feature, validation with real patients needed to be performed. This was carried out successfully, and full details of this can be found in Part 2 of this article, which will appear in an upcoming edition of The Hearing Review.
|
|
| Matthias Latzel, PhD, is clinical research manager, and Jennifer Appleton, MSc, is an audiology manager at Phonak headquarters in Stäfä, Switzerland. CORRESPONDENCE can be addressed to Dr Latzel at: [email protected] | |
References
1. Beard J, Nepomuceno H. Wind noise levels for an ITE hearing aid. Knowles Engineering Report No. 128, Revision A. Itasca, Ill: Knowles Electronics; 2001.
2. Brown DV, Mongeau LG. The design, construction, and validation of a small, low-speed, quiet wind tunnel with application to noise from the flow over a cavity. Internal Report 204: HL95–HL99. Herrick Laboratories, Purdue University; 1995.
3. Chung K, Mongeau L, McKibben N. Wind noise in hearing aids with directional and omnidirectional microphones: polar characteristics of behind-the-ear hearing aids. J Acoust Soc Am. 2009;125(4):2243-59.
4. Chung K, McKibben N, Mongeau L. Wind noise in hearing aids with directional and omnidirectional microphones: polar characteristics of behind-the-ear hearing aids. J Acoust Soc Am. 2010;127(4):2529-2542.
5. Chung K. Comparisons of first and second generations of wind noise reduction algorithms and effects of head hair. In press.
6. Kochkin S. MarkeTrak VIII: Consumer satisfaction with hearing aids is slowly increasing. Hear Jour. 2010;19:63.
7. E DIN EN 60 118 – 15. Akustik – Hörgeräte: Teil 15 Methoden zur Charakterisierung der Hörgeräte-Signalverarbeitung.
