A Perspective Post-Filtering for Microphone Array Beamforming’s Evaluation

Abstract

The use of microphone array (MA) is now widely popular, due to its ability of designed beampattern at certain location of target speaker and mitigating the influence of background noise field, interference and non-directional noise sources. MA beamformer has high spatial diversity, high directivity index for obtaining clean speech data. However, under complicated recorded situations, due to the effect of internal electrical noise field, the difference of started time recording between microphones, the error about sampling frequency, the miss alignment of configuration between microphones, the displacement of elements in MA systems, the effectiveness of beamformer often corrupted. In this distribution, the authors proposed an efficient post-Filtering for alleviating musical noise component and enhance speech intelligibility. Numerical results shown the mitigation of background noise level to 6.7 dB and improving the signal-to-noise (SNR) ratio from 7.6 to 10.9 dB.