Stability-controlled hybrid adaptive feedback cancellation scheme for hearing aids.

Area of Science:

• Hearing aid technology
• Acoustic signal processing
• Adaptive filter algorithms

Background:

• Adaptive feedback cancellation (AFC) is crucial for hearing aid devices (HADs) to increase stable gain.
• Open-fitting and in-ear HADs experience prevalent howling due to acoustic coupling, necessitating improved AFC.
• Existing AFC methods struggle with rapid recovery from feedback oscillations.

Purpose of the Study:

• To introduce a hybrid AFC (H-AFC) scheme for faster howling recovery in hearing aids.
• To enhance the performance of AFC by improving convergence rate and reducing misalignment.
• To maintain high speech and audio quality while minimizing feedback in HADs.

Main Methods:

• Proposed a hybrid AFC (H-AFC) scheme using a switched adaptive filter controlled by a stability detector.
• Implemented a soft-clipping-based stability detector to select between Normalized Least Mean Squares (NLMS) and Prediction-Error-Method NLMS (PEM-NLMS) algorithms.
• Evaluated the H-AFC scheme using speech and music signals, assessing convergence, misalignment, and added stable gain.

Main Results:

• The H-AFC scheme demonstrated significantly improved convergence and re-convergence rates, shortening howling duration.
• Achieved lower average misalignment and greater added stable gain compared to standalone NLMS or PEM-NLMS.
• Objective and subjective evaluations confirmed high-quality speech and music signal reproduction, outperforming the PEM-NLMS algorithm.

Conclusions:

• The proposed H-AFC scheme effectively reduces howling periods and improves overall hearing aid performance.
• This hybrid approach balances fast convergence with low bias, enhancing both stability and audio fidelity.
• The H-AFC system offers superior perceptual audio quality for hearing aid users compared to current state-of-the-art methods.