Abstract:In order to solve the problem that the speech signal is affected by reverberation in the closed environment, the improved Wiener filter based on LCMV is proposed. In this proposed algorithm, the frequency domain array signals can be obtained by the Fourier transform of the signals with the reverberation received by the microphone array and are processed according to frequency division method. The dividedfrequency signals also are processed by using linear constrained minimum variance beamforming filter method and according to different frequency reverberation time of each beam, and the beamforming algorithm is applied to the high and low frequency domain to improve the accuracy of reverberation suppression for the frequency domain array signals. The combination power spectrum estimation of the frequency domain array signal is used to treat Wiener postfilter in order to suppress the reverberation. Since the receivedreverberation signals of microphone array between the direct wave and reflected wave are not related, so we use space information of the receivedreverberation signals of microphone array to solve accurately the estimation problem of the Wiener filter. Finally, the time domain signal is obtained by the inverse Fourier transform. The test results show that the proposed algorithm has a significant improvement on the reverberation suppression, and the PESQ score of the speech enhancement system is improved by 0.26 under the condition of reverberation time 600 ms.