Energy losing rate estimation based microphone array speech enhancement

Based upon the techniques of delay and sum beamforming and Wiener filtering,an energy losing rate estimation based microphone array speech enhancement algorithm is proposed.By using a specific linear microphone array with different distance between sensors to monitor the energy losing before and after the beamforming performed on the embedded sub-array,the weight coefficients of Wiener filter are estimated and the objective of speech enhancement is achieved.Evaluations on a simulative corpus prove that comparing with the original speeches,the average improvements of the enhanced speeches on the SNR,log-spectral distance and perceptual quality are 17.1 dB,1.001 and 0.935 respectively,which shows a promising application prospect.