A master hearing aid system for auditory tests of hearing impaired people Stefan Stenfelt
A master hearing aid system for auditory tests of hearing impaired people
Stefan Stenfelt
Background
7 JUNE 2017 2 Stefan Stenfelt
• Investigations with participants with normal hearing function and hearing impairments.
• Need to compensate for audibility.
• Own hearing aids are ecological, but impose problems with different settings and lack of controlled gain
The N200 study
7 JUNE 2017 3 Stefan Stenfelt
N200 study, participants
7 JUNE 2017 4 Stefan Stenfelt
Three groups 1. Hearing impaired using hearing aids (n=200) 2. Normal hearing participants matched to group 1 (n=200) 3. Hearing impaired not using hearing aids (approx 100)
Group 1
N200 study, tests.
7 JUNE 2017 5 Stefan Stenfelt
Auditory tests • Pure tone thresholds, AC and BC • PB-words in quiet • DPOAE • Binaural TFS sensitivity (TFS-LF) • TEN HL • Spectro-Temporal Modulation
Cognitive tests • Phonological tests
• Semantic long-term memory
• Working memory
• Executive &inference-making functions
• General cognitive functioning
Outcome tests • HINT • Hagerman sentences • Samuelsson & Rönnberg sentences • Auditory inference-making test • Speech, Spatial and Qualities of hearing
(SSQ)
Hearing aid signal processing
7 JUNE 2017 6 Stefan Stenfelt
The Hagerman sentence test was conducted with three different amplification and signal processing 1. Linear amplification (voice aligned compression rationale) 2. Wide dynamic range compression (2:1 compression, 10 ms attack
and 80 ms release time) 3. Linear amplification + non-ideal binary mask noise reduction
Two different noises were used,
1. Stationary speech-weighted noise
2. 4-talker babble
In total 6 conditions
Controlled hearing amplification
7 JUNE 2017 7 Stefan Stenfelt
• Use hearing aids that are fitted to each individual
• Use a central hearing aid and provide the acoustic output to the participant (MOSAIC)
• Computer-based gain (MATLAB)
• Computer-based Master Hearing Aid System
Hearing aids fitted to individuals
7 JUNE 2017 8 Stefan Stenfelt
Pros
- Real acoustic setting
- Wearable
- Binaural
Cons
- No control of prescription implementation (even if generic)
- Acoustics not controlled (influence from head, pinna etc)
- Reprogramming, handling
- Restricted parameter values
Central hearing aid
7 JUNE 2017 9 Stefan Stenfelt
2cc coupler
Anechoic test-box
Amplifier
ER3 earphones
Pros
- Controlled acoustic input
- Easy programming
- No level problems
- Participant outside testroom
Cons
- No control of prescription implementation (even if generic)
- Restricted parameter values
- Noise
- Binaural problematic
Computer-based gain (MATLAB)
7 JUNE 2017 10 Stefan Stenfelt
Pros
- Controlled acoustic output (calibrated)
- Controlled gain function
- Signal processing
- Simulated environment
- Participant outside testroom
Cons
- Multiple sources not possible
- Primarily pre-recorded stimulus
Computer-based Master Hearing Aid System
7 JUNE 2017 11 Stefan Stenfelt
Pros
- Same as Computer-based gain system
- Real-time function
- Multiple sources
Cons
- Price!
- Non-wearable
Studies
7 JUNE 2017 12 Stefan Stenfelt
Hearing aid Central hearing aid Matlab based MHA
Speech perception
Speech perception Real life
Speech perception Simulated environment
Hearing aid evaluation
Signal processing development/evaluation
N200, results: Factor analysis
7 JUNE 2017 13 Stefan Stenfelt
Factor
SENSITIVITY
TEMPORAL
FINE STRUC
HE
AR
ING
PTA 0.91 0.05
TEN HL 0.06 0.16
DPOAE -0.29 0.03
TFS-LF -0.29 0.92
STM 0.10 0.48
PB words 0.37 0.34
COGNITION
CO
GN
ITIO
N Phonology 0.59
LT memory 0.53
WM 0.63
Executive func 0.92
General cog func 0.43
CONTEXT NO CONTEXT
OU
TC
OM
E HINT 0.87 0.32
Hagerman 0.25 0.83
S&R -0.52 0.08
AIM -0.05 0.40
SSQ -0.02 0.41
SENSITIVITY TEMPORAL CONTEXT NO
CONTEXT
COGNITION 0.13 0.30 0.16 0.37
SENSITIVITY 0.29 0.22 0.42
TEMPORAL 0.20 0.38
CONTEXT 0.33
Correlations between level 2 factors
N200, effect of signal processing
7 JUNE 2017 14 Stefan Stenfelt
SNR score, Hagerman sentences
Summary
7 JUNE 2017 15 Stefan Stenfelt
• For auditory testing, it is important to provide audibility of the signal
• Testing with wearable hearing aids resembles real life but lack
control of several parameters and have generally a limited set of
options
• A centrally placed hearing aid ensures a controlled gain but have
several drawbacks
• A computer-Matlab based gain control facilitates controlled, cheap
and easy setup. Cannot be used with live signals and is
problematic with several sound sources
• The Master Hearing Aid system have the advantages of the Matlab
based system with the addition of using live signals and multiple
sources. Slightly more complex programming/setup and expensive.
Applied Auditory Research, Linköping University
7 JUNE 2017 16 Stefan Stenfelt
www.liu.se
Stefan Stenfelt