Evaluation of feedback-reduction algorithms for hearing aids.

Area of Science:

• Audiology
• Digital Signal Processing
• Biomedical Engineering

Background:

• Hearing aid feedback, or whistling, is a common problem that limits the maximum usable gain.
• Adaptive feedback-reduction algorithms aim to increase stable gain and improve user satisfaction.

Purpose of the Study:

• To evaluate and compare the performance of three adaptive feedback-reduction algorithms in a digital hearing aid system.
• To assess the impact of these algorithms on maximum stable gain and subjective speech quality.

Main Methods:

• Implementation of three adaptive feedback-reduction algorithms: CNN (Closed-loop processing with No probe Noise), ONO (Open-loop with Noise when Oscillation detected), and ONQ (Open-loop with Noise when Quiet detected).
• Evaluation using dynamic feedback paths and hearing-impaired subjects, measuring maximum stable gain and subjective quality ratings.
• Analysis of algorithm performance across different hearing loss severities and vent conditions.

Main Results:

• The CNN algorithm achieved the highest average Added Stable Gain (ASG) of 8.5 dB, outperforming ONO and ONQ (5 dB ASG).
• Subjects with severe hearing loss experienced greater benefits, with the CNN algorithm yielding 13 dB ASG.
• Subjective speech quality ratings did not show significant differences, though probe noise from ONO/ONQ was disliked.

Conclusions:

• The continuously adapting CNN algorithm demonstrates superior performance for adaptive feedback reduction in digital hearing aids.
• CNN offers a promising solution for increasing stable gain, particularly for users with significant hearing impairment.
• Further research may be needed to address user objections to probe noise in intermittently adapting algorithms.