A Method for Reducing Speech Distortion in Minimum Variance Distortionless Response Beamformer

Speech processing devices such as mobile telephone, hearing aid, cochlear implant is commonly equipped with a designed microphone array (MA) to capture the acoustic environment. The MA signals often contain a mixture of desired speech component and unwanted surrounding noise (a combination of reverberation, different interferences, complex recording environment, third-party speaker). Minimum Distortionless Variance Response (MVDR) beamformer is one the most useful methods for reducing background noise and improving speech enhancement has been recently studied. MVDR beamformer merges the predefined constraints of response from a priori relative transfer function and the criteria of minimum total output noise power. Because of the microphone mismatches, the different microphone sensitivities, or the error of the direction of arrival of the interest signal, the output final signal of MVDR beamformer often corrupted. In this paper, the authors proposed an additive gain function for reducing speech distortion and enhancing MVDR beamformer's performance.