Speech enhancement using microphone array in moving vehicle environment

This paper proposes a robust speech enhancement method combining spectral subtraction and beamforming, which can be used as a preprocessor for speech recognition systems. Spectral subtraction is an effective method to reduce stationary additive noise from a single microphone signal. However, it has a major drawback, in that it introduces musical noise. In this paper, it is demonstrated that the proposed method improves existing spectral subtraction methods and reduces its residual noise using a microphone array.