1. Hearing Aids and Hearing Impairments
Meena Ramani
02/21/05

2. Dramatic Decrease In Audibility & Intelligibility
Original Speech
40dB conductive loss
P. Duchnowski and P. M. Zurek, “Villchur revisited: Another look at automatic gain control simulation of recruiting hearing loss,” J. Acoust. Soc. Am., vol. 98, no. 6, pp , Dec. 1995

3. Outline Facts on Hearing Loss Hearing Aids Cochlea-IHC and OHC
Presbycusis
Decreased Audibility
Decreased Frequency Resolution
Decreased Temporal resolution
Decreased Dynamic Range
Amplification Techniques
Linear
Compressive-Single/MultiBand

4. Facts on Hearing Loss in Adults
One in every ten (28 million) Americans has hearing loss and the prevalence of hearing loss increases with age.
While hearing aids can help about 95% (26 million) of them, only 6 million use hearing aids.
WHY?
Stigma associated with wearing a Hearing Aid (HA)
Denial about one’s Hearing Loss (HL)
Exorbitant cost (eg. A pair of Widex Senso Diva BTEs cost around $11,000)
Current HAs do not meet user expectations

5. Hearing Aids- An Engineering perspective
Area where vast improvement are possible
28 million Hearing Impaired people Huge Market($$$$)
Circuit design and Signal processing personnel
Circuit design:
Low power: 1.3V
Fast acting (delay < 10ms)
Small size
Lower cost
Signal Processing:
Biologically inspired/smarter algorithms
Restore all effects of hearing impairment.

6. Anatomy of a Hearing Aid
Microphone
Tone hook
Volume control
On/off switch
Battery compartment

7. Types of Hearing aids Behind The ear BTE In the Ear ITE
Completely in the canal
CIC
In the Canal
ITC

8. Cochlea-IHC and OHC Organ of corti: Inner Hair Cells (IHC)
IHC/OHC
3 times more OHC
Inner Hair Cells (IHC)
Afferent <to brain>
Outer Hair Cells (OHC)
Efferent <from brain>
Sharpen the traveling wave
Provide an amplification for soft sounds(40-50 dB SPL)
Damage in OHC/IHC Sensorineural Hearing Loss (SNHL)
Currently we don’t sharpen the peak of the traveling wave
OHCs are very mechanical..use a lot of oxygen…die first…mild-moderate loss..if IHC die then severe hearing loss
Presbycusis

9. Presbycusis Type of Sensorineural Hearing Loss
HL in aging ears; occurs due to damage in OHCs
Mild dBHL
Moderate   40-68 dBHL
Severe dBHL
Problems faced by people with presbycusis:
Decreased Audibility
Decreased Frequency Resolution
Decreased Temporal resolution
Decreased Dynamic Range

10. Decreased Audibility
90% of HI adults loose frequencies between 500Hz-4KHz
HF components of speech (consonants) are weaker than the LFs.
Loudness dominated by the LFs
“Speech is loud enough but not clear enough!”
To overcome this:
HA has to provide more gain at HFs.

11. Decreased Frequency Resolution
Asymmetry of traveling wave
Eg. Reverse Audiogram
OHCs do not sharpen the traveling wave.
Decreases the ability to distinguish close frequencies
Upward spread of masking low frequencies mask more than high frequencies
Normals and HI: Poor resolution at high intensities
To overcome this:
HAs less gain at LFs
Try to remove noise before entering HA. Beamforming

12. Decreased Temporal Resolution
Intense sounds mask weaker sounds that immediately follow them.
To overcome this:
Fast acting compression
<Problem: Changes the speech cues; decreases intelligibility though it increases audibility!>

13. Dynamic Range of Hearing
The practical dynamic range could be said to be from the threshold of hearing to the threshold of pain
Sound level measurements in decibels are generally referenced to a standard threshold of hearing at 1000 Hz for the human ear which can be stated in terms of sound intensity:
Equal Loudness Contours

14. Decreased Dynamic Range/Recruitment
SNHL increases threshold of hearing much more than the threshold of pain; thus decreases the Dynamic Range of the ear.
To overcome this:
HA has to provide Compression; cut down amplification as sound gets louder.

15. Decreased Dynamic Range/Recruitment
Figure 7.1. Typical loudness growth functions for a normal-hearing person (solid line) and a hearing-impaired person (dashed line). The abscissa is the sound pressure level of a narrowband sound and the ordinate is the loudness category applied to the signal. VS, very soft; S, soft; C, comfortable; L, loud; VL, very loud; TL, too loud.
Figure 7.2. The response of a healthy basilar membrane (solid line) and one with deadened outer hair cells (dashed line) to best-frequency tone at different sound pressure levels (replotted from Ruggero and Rich 1991).The slope reduction in the mid-level region of the solid line indicates compression; this compression is lost in the response of the damaged cochlea.

16. Linear Amplification
Figure 7.3. Loudness growth functions for a normal-hearing listener (solid line), a hearing-impaired listener wearing a linear hearing aid (short dashed line), and a hearing-impaired listener wearing a compression hearing aid (long dashed line with symbol).
HA wearer adjusts gain, using volume control, as the level of environment changes.

17. Compressive Amplification
Slope=1/Compression Ratio
Imitates compression carried out by OHCs
Fast Acting/Syllabic Compression:
Attack time~5ms
Release time~60ms
Choose release time
To avoid distortion
To normalize loudness from phoneme to adjacent phoneme for syllabic compression
Figure 7.4. Typical input-output function of a compression hearing aid measured with a pure tone stimulus at multiple levels. The function depicted shows linear operation at low and high input levels, and 3 : 1 compression at mid-levels. Different compression hearing aids have different compression ratios and different levels over which compression occurs.

18. Time Constants: Overshoot and Undershoot
Affects Intelligibility
Makes consonants be identified as plosives
Reduce effects:
Clipping overshoot
Delaying gain
Undershoot:
When release time isn't that large, then forward masking lowers the affect of undershoot
Figure 7.5. A demonstration of the dynamic behavior of a compressor. Top: Level of the input signal Middle: Gain that will be applied to the input signal for 3 : 1 compression, incorporating the dynamics of the attack and release time constants. Bottom: The level of the output signal, demonstrating overshoot (at 0.05 second) and undershoot (at 0.15 second).

19. Singleband/Wideband Adjust gain across all frequencies equally
Preserves spectral shape over short time scales; speech cues
Choose gain based on highest level; Spectral peak
Speech has multiple spectral peaks. Inadequate selection of gains.
Figure 7.7. Amount of compression applied to music by a wideband compressor (squares) and a multiband compressor (circles).The compression was measured by comparing the peak/root mean square (rms) ratio of the music into and out of the compressor over different frequency regions. The open symbols on the left show the compression ratio calculated from the change to the broadband peak/rms ratio. The filled symbols show the change to the peak/rms ratio in localized frequency
regions.

20. Multiband Compressor
Normally upto 20 bands are used with varying compression ratio per band.
Adjust gain/compression in each band independent from other
Change in spectral contrast across bands may cause perceptual consequences though it restores normal loudness.
STI<Speech Intelligibility Measure> of compressed speech does not correlate to Listening tests.
With more experience people who use multiband HAs get adjusted to the change in spectral shape/cues.
Typically vowel perception is not affected as much as consonant perception
Overamplification occurs at crossover between bands<To avoid this increase overlap between bands so that gain at a frequency is controlled by more than 2 bands>