An improved MVDR beamformer’s noise reduction based on post- filtering

Abstract

Nowadays, microphone array (MA) technology has been installed into almost acoustic equipments, such as surveillance device, mobile phone, voice con trolled- device, teleconference system. Due to its high diversity, high directivity index towards the sound source, MA allows obtaining the original useful of desired talker, the clean speech data or necessary component while attenuating the effect of interferences, annoying background noise, third- party talker with promising spatial diversity, directivity index or different acoustic factors. Minimum variance distortionless response (MVDR) beamformer based on the optimum constrained formulation of mitigating complicated environment, back ground noise field while saving the speech component. However, in realistic recording scenario, because of the error of estimating the direction of arrival, the different microphone quality, the error of sampling frequency, the displacement of microphone geometry, the existence of internal electrical noise, the MVDR beamformer’s evaluation degraded. Consequently, the existence of the remaining noise or speech distortion usually occur. This contribution introduced an additive post- Filtering, which blocks the noise level and increases the speech quality from 10.8 to 12.7 dB and reduces the noise level to 11.2 dB. The numerical experiment has confirmed the effectiveness of the author’s suggested technique in improving the robustness of MVDR beamformer in complex environment.