Chapter 3 - Modified least mean square adaptive filter for speech enhancement

Abstract

Speech signals play a major role in the human communication system. In speech processing applications like hearing aids, automatic speech recognition, mobile phones, car communication, voice coders, teleconference systems, hands-free phones, forensics, and so on, the elimination of background noise is a tedious process. Noise reduction algorithms for speech enhancement are used to overcome this problem in speech processing applications. This type of speech enhancement algorithm eliminates background noise and improves speech signal intelligibility and quality of speech. In this chapter, we develop a Modified Least Mean Square Adaptive Noise Reduction (LMS-ANR) algorithm for enhancing the Tamil speech signal with acceptable quality under nonstationary noise environments. This proposed algorithm automatically adapts its coefficients with respect to input noisy signals.