Improving the quality of speech signal processing via a microphone array

The topic of the article is about modeling broadband acoustic signal processing via a microphone array. The article solves the problem of tuning the microphone array to suppress the signal from the source at one angle and to receive the signal undistorted from the source at the other angle during the modeling. Weighted summation in the frequency domain is used; threshold limits are imposed on weight coefficients in order to minimize distortion of the useful signal during processing. The optimal threshold values of weighting coefficients have been determined according to the results of expert statistical tests. Setting these threshold values, we achieve the conditions when both the artifacts (hissing, whistling, ringing, clanging) and the wanted signal distortions (change in the voice timbre, muting consonants) are minimally noticeable.