A Coherence - Based Post-Filtering for Minimum Variance Distortionless Response Beamformer

At present, the performance of traditional digital signal processing for mobile telecommunications, teleconference and processing speech often degrades due to many reasons, such as the non-stationary surrounding noise, the presence of third-party speakers, or the lack of installed equipment’s limitations. Therefore, the implementation of microphone array technology deal the above problem by exploiting the spatial beampattern or a priori information about the distributed of microphone array, the direction of arrival of interest signal, the characteristics of background noise to achieve the advantage of noise reduction while saving the target talker without speech distortion. In this article, the authors proposed an useful Post-Filtering to suppress the remaining noise at the output of MVDR beamformer’s output to enhance the overall evaluation of MVDR beamformer, increase the speech quality in the term of the signal-to-noise ratio (SNR). The illustrated experiment shows that the proposed post-filtering has the capability of eliminating annoying noise while saving the original speech component.