Adaptive Noise Cancellation Algorithm for Speech Processing

Abstract

LMS adaptive noise cancellers are often used to recover signal corrupted by additive noise. A major drawback of conventional LMS algorithms is that the excess mean-square errors increase linearly with the desired signal power. This results in degraded performance when the desired signal exhibits large power fluctuations. In this paper, a normalized difference LMS (NDLMS) algorithm is proposed to deal with the situation when the desired signal is strong, e.g., speech signals. Simulations were carried out using real speech signal with different noise power levels in both stationary and nonstationary noise environments. Results demonstrate the superiority of the proposed NDLMS algorithm over conventional LMS algorithms in achieving much smaller steady-state excess mean square errors.