Microphone Array Post-Filter Based on Auditory Filtering

In this paper, an auditory filtering based microphone array post-filter is proposed to enhance the quality of the output signal. By using a gammatone filterbank to band pass each input of the array, the input signals are decomposed into a two-dimensional T-F representation. Then, for each auditory filter channel, the post-filter´s coefficients are estimated in each frame using the decomposed multi-channel input signals. Followed by the post-filtering and synthesis processing, the enhanced speech with better quality is acquired. Systematical evaluations on the CMU microphone array database prove that the proposed method could improve not only the noise reduction measure but also the speech quality measures.