A MAP-based speech enhancer combined with two-channel beamformer in automobile environment

In single microphone speech enhancement, a MAP-based speech enhancer is known as a very effective method. Although the MAP-based speech enhancer has a strong noise reduction capability, it also provides speech quality degradation in a low SNR environment such as in an automobile. To solve this problem, we improve the SNR by using null beamformer before applying the MAP-based speech enhancer. However, the suppressed signal includes low frequency speech components. Then we also apply the MAP-based speech enhancer to the signal suppressed by the null beamformer, and we sum up the two enhanced speech signals. In this paper, a novel approach is proposed for two-channels speech enhancement which is based on the MAP-based speech enhancer and null beamformer. From computer simulations, we confirm the effectiveness of the proposed method.