An improved NLMS algorithm based on speech enhancement

Abstract

Currently, adaptive noise cancellation algorithm is the mainstream technology in the field of noise cancellation, many research schemes are put forward on the basis of it. Least mean square (LMS) adaptive algorithm is the first one to be studied, there are many modified algorithm based on LMS, such as Normalized Least Mean Square (NLMS), Proportionate NLMS (PNLMS), etc. All kinds of adaptive algorithms have their own advantages and disadvantages, and their corresponding improvement schemes are different. With increasing requirement from adaptive noise cancellation in practical applications, these algorithms are constantly growing and improving. By studying the principles of adaptive noise cancellation (ANC), this article uses the regularization NLMS algorithm to improve its application in noise cancellation system and compares the simulation results of some performance with traditional NLMS algorithm. In speech signal processing, both methods can suppress interference signals to extract useful signals from the high background noise. However, the regularization NLMS algorithm can exhibit a faster convergence, a greater stability and a better ability to suppress interference.