Electroacoustic Testing of DSP Hearing Aids

1
Electroacoustic Testing of DSP Hearing Aids

• Christine Cameron Mary Hostler
• MCHAS Team
• University of Manchester

2
Introduction

• Testing DSP hearing aids guidelines prepared
  for the MCHAS project
• People it will involve Teachers of the Deaf and
  Educational Audiologists
• working in the field

3
Overview

• Change is necessary.
• What are the options available?
• What are the best solutions?

ICRA?
Test signals
Settings?
4

• Regular electroacoustic testing is still
• important
• Test boxes currently in use are less
• than ideal for testing complex hearing
• aids with multichannels and
• compression.
• ToDs and Ed Auds cannot change or
• read hearing aid parameters without
• NOAH, manufacturers software,
• programming cables and hi-pro box.

5

• Full specification checks are still
• desirable at clinic based hearing aid
• reviews, or if a hearing aid is found on
  routine testing to deviate from usual
  performance. (/- 3dB from Baseline FRC, or THD
  unacceptable)
• Do this in Test mode in h.a. software

6
Guidelines for Testing DSP Hearing Aids

• Note how the hearing aids have been set
• Feedback management
• Memory 2
• Directional / Omnidirectional mic
• Noise reduction

7
(No Transcript)
8
(No Transcript)
9
2. Carry out a subjective listening check
10
3. Generate baseline FRCs / Run FRCs at three
input levels

• Input signal
• Pure Tones
• Speech shaped noise
• Pulsed or modulated speech shaped signal

50dB SPL
65dB SPL
80dB SPL
Test box
11
FRCs for Digifocus II
12
4. Compare curves obtained with baseline FRCs,
they should be within / - 3dB 5. Check THD
using pure tone sweep. This should be lt10 for
the 65dB SPL input curve 6. Troubleshoot. Arrange
full specification check at clinic if necessary
13
Issues raised by Step 1

• 1st Major issue is that
• INFORMATION SHARING
• between Audiology and Education
• IS CRUCIAL

14
Issues raised by Step 1

• Any changes to the programming require
• new baseline FRCs
• Frequency of generating new baseline FRCs
  may increase
• fine tuning taking place
• When new earmoulds fitted new RECDs should be
  measured and adjustments made to the hearing aid
  to ensure match to target is still as close as
  possible
• Gradual introduction of features

15
Issues raised by Step 1

• Feedback management
• if this has been activated at fitting, or at
  subsequent clinic appointments, it will affect
• the baseline FRCs e.g. presence of a notch or
  high frequency gain reduction. (For most of the
  current NHS DSP aids)
• The testing - if pure tones are being used and
  the aid has active feedback management or
  suppression. (For eg Danalogic 283D, Supero 413,
  Prisma 2DSP)
• demo 1
• Feedback management notch

16
(No Transcript)
17
Issues raised by Step 1

• What is contained in Memory 2 is important to
  note. If it contains a different program, two
  sets of baseline FRCs will be needed they may
  be very different.
• demo 2
• Normal use program Vs noisy environment
• COMPARE LIKE WITH LIKE!!

18
Program 1
Program 2
19
Issues raised by Step 1

• If the directional microphone facility has
  been activated (usually in memory 2 but could
  be on function switch) this may have an effect on
  test results.
• demo 3
• Directional mic Vs Omnidirectional mic

20
(No Transcript)
21
Issues raised by Step 1

• If a Noise Reduction feature has been
  activated, then the INPUT SIGNAL in use can have
  a particularly dramatic effect on test results.
• demo 4
• DSP hearing aid with Noise Reduction feature on
  - FFT (speech noise) stimulus
• FFT Fast Fourier Transform

22
(No Transcript)
23
Issues raised by Step 2

• May need attenuators for your stetoclip (and
  parents stetoclips) as the DSP aid may have no
  Volume Control or not a great deal of adjustment
  on the VC
• Take care when listening!!

24
Care when listening
Use attenuators or VC in the stetoclip
especially if there is no volume control on the
hearing aid
25
Issues raised by Step 3

• 2nd Major issue is the
• INPUT SIGNAL USED TO TEST DSP HEARING AIDS

26
Issues raised by Step 3

• Input Signal has been shown to have a big
  effect on test results when DSP aids have Noise
  Reduction features on. (eg Danalogic 283 D)
• It is also an issue for ALL complex hearing
  aids with multichannels and compression
• demo 5
• blooming effect

27
Issues raised by Step 3

• Ideally, speech or a speech shaped signal which
  is pulsed or modulated should be used but most
  Services for H.I. dont have this signal
  availableyettest boxes for Education are
  available via the project.
• Next best is speech shaped noise (such as the
  composite signal on FONIX FP 40 test boxes)
• Worst is pure tones various possible funny
  effects!!! but STILL MUCH BETTER THAN NOTHING
  IF YOU ARE AWARE OF THESE EFFECTS.

28
Issues raised by Step 3

• Possible effects with pure tonesto recapare
• Filters followed by compressors in multichannel
  hearing aids may result in the compressor undoing
  the effect of the filter so the FRC looks
  different from one measured with a broadband
  speech noise or digital test signal i.e.
  Blooming effect low frequency increase, and
  high frequency decrease
• Spikey FRC if there is active DFS
• Drop in gain / output if Noise Reduction feature
  is on.

29
Issues raised by Step 3

• When using broadband, noise like signals,
  where all the frequencies are present at once,
• GAIN rather than OUTPUT must be measured
• demo 6
• Output Vs gain with FFT speech noise
• THD cannot be tested (only possible to do this
  with pure tones)
• Pure tones must be used for estimating OSPL 90

30
Step 3 demonstration

• demo 7
• Three baseline FRCs obtained with speech
  weighted noise (using GAIN /Frequency response
  curves) These curves show the nonlinear
  functioning of the hearing aid, with different
  gain applied for different input levels.
• Input/output graphs will also show the nonlinear
  function and are worth running at 500Hz and 2kHz.

31
STEP 3. Generate baseline FRCs
Input signal Pure Tones Speech shaped
noise Pulsed or modulated speech shaped signal

• Non-linear hearing aids provide different amounts
  of gain for different input levels.
• Generate baseline curves at 50, 65, 80 dB.

32
So far..

• These guidelines have been in use by First Wave
  Second wave sites
• Every site in 1st wave studies had some hearing
  aids fitted as WDRC but which were in fact
  functioning linearly it is important to know
  how the aids are functioning for counselling the
  user and for setting up FM systems appropriately.
  
• Our definition of WDRC that the kneepoint is
  low, lt 60 dB SPL at either 500Hz or 2kHz. CR
  typically is low, lt51.
• 50 dB input FRC difficult to obtain if noisy test
  conditions in some schools.

33
Conclusion

• Blooming effects!!!
• Sounds more problematic than it is!!
• Worth extra hassle if children benefit
• CONTINUE TO TEST DSP AIDS !!

34
Thank You!!