Efficient Modeling of Acoustic Feedback Path in Hearing Aids by Voice Activity Detector-Supervised Multiple Noise Injections.

Abstract

Adaptive Feedback Cancellation (AFC) techniques are widely used to eliminate the undesired acoustic feedback effect arising in the Hearing Aid Devices (HADs) due to the coupling between the speaker and the microphone of the HAD. This paper proposes a method to eliminate the acoustic feedback effect in the HADs in presence of noisy environment. The method involves utilization of a computationally efficient Spectral Flux feature-based voice activity detector (VAD), which controls the process of Noise Injection in the proposed AFC algorithm (SFNIAFC). The proposed algorithm's performance is objectively evaluated using Misalignment (MISA) and Perceptual Evaluation of Speech Quality (PESQ) criteria for realistic noisy conditions. The simulations performed for the proposed method shows faster convergence and reduction in the MISA values with high PESQ values in comparison to the earlier method. Subjective test results support the effectiveness and better performance of the proposed algorithm for the HAD applications over earlier method.