Robust adaptive microphone array processing for hearing aids: realistic speech enhancement

Area of Science:

• Acoustics
• Signal Processing
• Hearing Aid Technology

Background:

• Hearing aids traditionally use single microphones, limiting performance in noisy environments.
• Combining multiple microphone outputs (microphone arrays) offers potential for improved speech intelligibility.
• Predicting realistic performance gains is crucial for developing effective hearing aid technologies.

Purpose of the Study:

• To investigate combining microphone array outputs for hearing aid input.
• To conservatively predict achievable performance gains of microphone arrays over single microphones.
• To evaluate array performance using speech reception threshold (SRT) and signal-to-noise ratio (SNR) changes.

Main Methods:

• Performance improvement measured by change in SRT between single and multi-microphone conditions.
• Utilized intelligibility averaged gain and signal-to-noise ratios (SNRs) to evaluate array performance.
• Investigated effects of acoustic headshadow, room reverberation, and placement uncertainties.

Main Results:

• Predictions using intelligibility averaged gain correlate well with actual SRT changes.
• Robust adaptive microphone array processors offer realistic speech enhancement.
• Performance improvements demonstrated for fixed, block processed, and adaptive spatial filters.

Conclusions:

• Microphone arrays provide significant, realistic speech enhancement for hearing aids.
• Robust block processed array performance is achievable with adaptive implementations.
• Beamformer design criteria limit desired signal cancellation, ensuring robust performance.