The Using of Real Part Component to Enhance Performance of MVDR Beamformer

The benefit of using of microphone array is utilizing of spectral domain, observed diversity, time domain, which contained in a received array signals to compute an useful signal of interest while eliminating background noise or estimate an necessaries parameters, that required for further signal processing. Dual - microphone system is one of the most widely integrated in almost speech application, cause it’s compact and it’s convenient to implement method signal processing, which based on microphone array. The basic algorithm is MVDR filter, that aims of minimizing the total output noise power while ensuring saving the target speaker. Unfortunately, many objective reason affect on performance of MVDR filter; microphone mismatch, the different sensitivities of microphone, the unknown exactly of direction of arrival also degrade the effectiveness of MVDR beamformer. In this paper, the authors considered the problem of speech distortion at the output’s beamformer. The authors proposed a modified matrix cross - spectral, that uses real part component to enhance the performance of MVDR Beamformer, and reducing the speech distortion.