Considerations for the Development and Fitting of Hearing-Aids for Auditory-Visual Communication Ken W. Grant and Brian E. Walden Walter Reed Army Medical.

Considerations for the Development and Fitting of Hearing-Aids for Auditory-Visual Communication

Ken W. Grant and Brian E. Walden

Walter Reed Army Medical Center

Army Audiology and Speech Center

Washington, DC 20307-5001

Typical Listening Environments for Multi-Memory Hearing Aids

• Quiet

• Background Noise– Low-frequency– High-frequency– Multiple Talkers

• Reverberation

• Music and Environmental Sounds

• Face-to-face communication is the most common of all listening environments.

• Should hearing aids be programmed differently when visual speech cues are available?

Auditory-Visual Listening Environments

• If we improve auditory-only speech recognition, do we necessarily improve auditory-visual speech recognition?

• What speech information is provided by speechreading?

• What speech information is provided by hearing aids?

• To what extent is the information provided by speechreading and by hearing aids redundant?

• What frequency regions best convey information that is complementary to speechreading?

If we improve auditory-only speech recognition, do we

necessarily improve auditory-visual speech recognition?

• Recognition of medial consonants (/C/) spoken by a female talker and recorded on optical disk.

• Manipulated the auditory intelligibility by band-pass filtering.

• Compared A and AV speech recognition scores for normal-hearing subjects.

70

80

90

100

30 40 50 60 70 80 90 100

Auditory Consonant Recognition (%)

r = 0.38

Au

di t

or y

- Vis

ual

Co

nso

na

nt

Re

cog

ni t

i on

(%

)

What speech information is provided by speechreading?

• Recognition of medial consonants (/C/) spoken by female talker and recorded on optical disk.

• Speechreading only.

• Measured percent information transmission of voicing, manner, and place cues.

0% 3%4%Voicing

Manner

Place

Other

Visual Feature Distribution%Information Transmitted re: Total Information Received

93%

HYPOTHESIS: The amount of benefit obtained from the combination

of visual and auditory speech cues depends of the degree of redundancy between the two modalities.

• Speechreading provides information primarily about place-of-articulation.

• Hearing aids that provide primarily (redundant) place information will result in small AV benefit.

• Hearing aids that provide (complementary) voicing and manner information will result in large AV benefit.

0

10

20

30

40

50

60

60 70 80 90 100Auditory Voicing+Manner Information (re: Information Received)

0

10

20

30

40

50

0 5 10 15 20 25 30Auditory Place Information (re: Information Received)

r = -0.87

r = 0.88

AV

Ben

efit

(A

V-A

)A

V B

enef

it (

AV

-A)

What speech information is provided by hearing aids?

• Twenty-five patients fit with the ReSound BT2 multi-band wide dynamic range compression hearing system.

• Recognition of medial consonants (/C/) under four receiving conditions (speech level at 50 dB SPL):– Unaided Listening (without hearing aid or visual cues)– Aided Listening (with hearing aid, no visual cues)– Unaided Speechreading (without hearing aid)– Aided Speechreading (with hearing aid)

E

E

E E

EE

EE

E E

I

I

II

II

II I I

100

90

80

70

60

50

40

30

20

10

0

100 1000 10000

Frequency (Hz)

E Right (N=25)

I Left (N=25)

NU-6Right: M=86.1% (sd: 6.7%)Left: M=85.1% (sd: 6.8%)

Hea

rin

g T

hre

sho

ld (

dB

HL

)

0102030405060708090

100

Voicing

Manner

Place

0102030405060708090

100

0102030405060708090

100

0 10 20 30 40 50 60 70 80 90 100Consonant Recognition Without Hearing Aid (%)

Listening With Hearing Aid

Listening With Speechreading

Hearing Aid Plus Speechreading

To what extent is the information provided by speechreading and by hearing aids redundant?

• Amplification and speechreading provide somewhat redundant information.

– Hearing aid provided information primarily about place-of-articulation. Smaller gains over unaided hearing were achieved for voicing and manner cues.

– Speechreading provided substantial improvement over unaided hearing for place and some improvement for manner. No benefit for voicing.

What frequency regions best convey information that is complementary to speechreading?

• Auditory recognition of medial consonants (/C/) by normal hearing subjects.

• Band-pass filtered speech conditions with equal Articulation Index.

• Analyzed confusions for information transmission of voicing, manner, and place features.

0

10

20

30

40

50

60

70

80

90

100

200 1000 6000

Center Frequency (Hz)

Place

Manner

Voicing

V

M

P

AI=0.1

Per

cen

t In

form

atio

n T

ran

smit

ted

0

10

20

30

40

50

60

70

80

90

100

200 1000 6000

Center Frequency (Hz)

Place

Manner

Voicing

V

M

P

AI=0.2

Per

cen

t In

form

atio

n T

ran

smit

ted

0

10

20

30

40

50

60

70

80

90

100

200 1000 6000

Center Frequency (Hz)

Place

Manner

Voicing

V

M

P

AI=0.3

Per

cen

t In

form

atio

n T

ran

smit

ted

SUMMARY

• Improving auditory speech recognition does not necessarily improve AV speech recognition.

• To improve AV speech recognition A and V cues should be maximally complementary.

• Speechreading provides information about place-of-articulation.

• Hearing aids tend to provide mostly place information, making them somewhat redundant with speechreading.

• Complementary cues to speechreading (voicing and manner) are best conveyed by low-frequencies.

Recommendations for designing hearing aids for Auditory-Visual speech communication

• Programming should focus on improving the recognition of voicing information, and to a lessor extent, manner-of-articulation information.

• Since voicing and manner information are primarily low frequency information, extend frequency response to include this region.

• May need to consider effects of compression on low-frequency amplitude envelope.

• Traditional concerns about upward spread of masking may not be warranted under auditory-visual conditions.

Acknowledgement

• NIH Grant DC00792

• ReSound Corporation, Redwood City, CA

• Department of Clinical Investigation, Walter Reed Army Medical Center, Washington, DC