Client adjustment of amplification Is this the future

1
Client adjustment of amplification Is this the
future?

• Gitte Keidser, Wouter Dreschler, Elizabeth
  Convery, and Harvey Dillon
• National Acoustic Laboratories and
• Co-operative Research Centre for
• Cochlear Implant and Hearing Aid Innovation
• Academic Medical Center, Amsterdam

2
State-of-the-art hearing aids

• The technology features
• Multi-channel and non-linear processing
• Noise reduction/speech enhancement
• Adaptive directionality
• The acoustical environment is taken into account
• Environment classification
• Multi-programs (user-driven)
• The number of fitting parameters is increasing
  faster than we can develop generic procedures to
  prescribe them!

3
Fitting alternatives

• Proprietary device specific fitting algorithms
• Trial and error
• Use trial periods and complaint-driven fine
  tuning
• The application of a set of background noises
• Usually with sound video (e.g. Beltone Avenue/
  Amplifit)
• Client adjustments using an interactive or
  adaptive procedure
• Client adjustments in real-life environments
• Trainable hearing aid

4
Research questions

• How does different simple control configurations
  for adjusting the frequency gain response affect
  the adjustments?
• How reliably can a client select the preferred
  amplification curves with each control
  configuration?

5
Methods

• 24 adult hearing aid users
• 7 flat loss, 10 gently sloping loss, 7 steeply
  sloping loss
• 6 real-life sounds, presented as video signals

Keidser et al, NAL
6
Methods

• Four different controllers

7
(No Transcript)
8
Methods

• Amplification
• NAL-RP (each individual)
• fixed 21 compression (each stimulus)
• roving of overall gain and frequency response
  slope re NAL-RP (each test condition)
• Two adjustments

9
Parameters

• Number of key presses (total and per key)
• Time needed for the preference setting
• Preferred amplification curves
• Subjective judgements of each controller
  (questionnaires)

10
Results the speed of adjustment

• Controllers B and C appear to be slower
• B uses volume/slope/contrast
• C uses bass/mid/treble
• Controller D (volume/bass/ treble) is almost as
  fast as controller A (volume/slope)
• There are some some differences between videos
• There is a trend that non-speech videos need less
  time and fewer key actions

11
Results subjective comments about the controllers
Controllers A (volume/slope) and D
(volume/bass/treble) are regarded as the easiest
to use
12
Results subjective comments about the controllers
Controller C (bass/mid/treble) is regarded as too
subtle
13

• Controllers A (volume/slope) and D
  (volume/bass/treble) are clearly preferred
• Please note that D is also a 3-key controller
• Almost all subjects like the idea of using their
  (preferred) controller with their own hearing
  aids

14
Results final setting
400 1250 4000
400 1250 4000
400 1250 4000
400 1250 4000
Reasonable similarities between the overall
results for different controllers
15
Results reproducibility parametersper
controller and per video fragment

• The average test-retest SD is 2.9 dB
• Controller B (volume/slope/contrast) shows worst
  reproducibility
• There is no large effect of the type of video
• For all subjects the starting position has a
  clear effect on the final preference.

16
Results the effect of starting point per subject
Gain _at_ 400 Hz Gain _at_ 1250 Hz
Gain _at_ 4000 Hz

• For all subjects clear differences !!
• Start at steeper than NAL gives the steepest
  amplification curves
• The overall gain is not affected by the baseline
  start

17
Conclusions

• Some differences between controllers
• Controllers A (volume/slope) and D (volume/bass/
  treble) were the fastest
• Controller A (volume/slope) is regarded the least
  subtle
• There is a clear subjective preference for
  controllers A (volume/slope) and D
  (volume/bass/treble)
• The end results do not strongly depend on
  controller style
• There is a reasonable precision in the responses
• Average test-retest SD 2.9 dB
• The starting baseline has a very high impact on
  the selected amplification curve

Keidser et al, NAL
18
Thank you for listening