Reverberation suppression algorithm for LS beamforming based on Wiener post-filter

Abstract

Reverberation causes speech signal time and frequency to blur, and speech is difficult to understand when serious. In order to improve the speech quality, this paper presents a least squares beamforming reverberation suppression algorithm with Wiener post-filtering. The algorithm divides the reverberated speech signal into direct part and reverberation part, and obtains the gain estimation of Wiener post-filter. For the low-frequency part of the speech signal, the noise coherence is strong, and the least squares beamforming algorithm is used to divide the frequency. And finally the optimal weight is gotten. Simulation results show that the algorithm has good reverberation suppression.

Keywords

• beamforming
• least squares
• reverberation
• Wiener post filtering