Hard Thresholding Algorithm for Adaptive Feedback Cancellation in Single Microphone Hearing Aids

Assistive listening devices (ALDs) are often challenged with acoustic feedback issues. To address this issue, adaptive feedback cancellation (AFC) estimation algorithms have undergone numerous advancements over the years. Typically, the acoustic feedback path exhibits sparsity, which can be exploited in AFC to improve their performance. In this paper, we show that the performance of an AFC algorithm can be improved in terms of faster convergence speed, lower steady state, and better short-time intelligence (STOI) score using the hard thresholding (HT) approach borrowed from the compressive sensing (CS) literature.