A practical acoustic echo cancellation using normalized least mean square adaptive filters

Abstract

Acoustic echo arise as a development of unorganized reflection during propagation. The study developed an acoustic echo removal algorithm, capable of providing results for personal computer application such as adaptive filtering techniques for reduction of unwanted echo and to increasing communication quality. The research deplored MATLAB software application for the investigation of echo canceller using new normalized least-mean squares (NLMS) adaptive filter, and a double-talk detector (DTD). Adaptive filter estimates the echo path based on a replica of the echo created and subtracted from the combination of the actual echo and the near-end speech signal, which is then used to negate the echo in the return signal. The study unveiled that the better the adaptive filter emulates the echo, the more successful the cancellation. The task of the doubletalk detector is to sense the doubletalk and to stop the adaptive filter in order to avoid divergence.

Keywords: Filters, NLMS, Echo cancellation, DTD