Performance improvement of acoustic feedback cancellation in hearing aids using linear prediction

Acoustic feedback cancellation (AFC) is required to mitigate the howling effect in hearing aids due to the acoustic feedback problem. In this paper, a technique is proposed to improve the performance of AFC in the hearing-aid devices. Due to the existence of the speech input energy within the error signal, the adaptation of the adaptive filter in the AFC system is therefore severely disturbed, and results in misconvergence of the adaptive filter. The proposed technique based on a forward linear predictor is employed so that the predicted speech input can be removed from the error signal, and thus, more accurate estimation of the acoustic feedback signal can be obtained. Computer simulation based on real speech signals demonstrates the effectiveness of the proposed technique.