Wiener Filter - based on Microphone Array Beamforming in Adverse Environment

Abstract

Nowadays, the using of microphone array (MA) beamforming is common widely integrated into numerous speech applications, due to its convenience of steering the designed beampattern toward the sound source location while attenuating interference and unwanted noise from other directions. MA beamformer utilizes the prior spatial information, the properties of surrounding situation, the characteristic of observed MA signals and different MA features for obtaining noise reduction and extracting desired target speaker without distortion at same time. In this article, the authors proposed an efficient modified Wiener filter for enhancing MA beamforming. The illustrated experiment was conducted in realistic recording environment to demonstrate the capability of saving clean speech data while suppressing surrounding noise, third-party talker and interference. The promising result has confirmed in removing residual noise level to 10.2 (dB) and increase speech quality in the term of signal-to-noise (SNR) from 9.6 to 10.9 (dB). The author’s approach owns the flexible ability for addressing other complicated problems.