A Robust Speech Enhancement Method Based on Microphone Array

In this paper, a robust noise suppression algorithm is introduced by combining beamforming technique and multiband spectral subtraction. Robust speech enhancement algorithms are critical to improving the performance of speech recognition systems. Basic spectral subtraction performs effectively for suppressing additive stationary noise received from a single microphone, but it will also introduce musical noise into our enhanced signal. In order to resolve this shortcoming, the proposed algorithm in this paper improves the existing spectral subtraction algorithm with a microphone array to suppress its residual musical noise. Results show that, the proposed algorithm works effectively on reducing both background noise and musical noise.