一种适用于混响环境的麦克风阵列语音增强方法

A Microphone Array Speech Enhancement Method for Dereverberation

在小型视频会议系统中,影响语音处理系统性能的主要因素是房间混响效应。现有语音增强方法大都采用加性噪声模型,没有考虑混响;而基于倒谱的语音去混响方法大多计算复杂。为此,本文给出了一种改进的麦克风阵列倒谱域语音去混响方法。该方法首先计算含噪语音倒谱域的最小相位分量,并对其进行波束形成,再进行低通滤波;然后利用人耳对相位信息的不敏感性,采用含噪语音的相位来合成增强后的语音信号,从而有效地降低了运算量。仿真结果表明,该方法在混响环境下能有效地改善语音质量。

The speech processing system is mainly affected by the reverberation for the small video meeting room. But the present speech enhancement methods mostly use additive noise model which doesn＇t take reverberation into accountant. And the speech der- eververation methods based on cepstrum is usually complicated. So this paper presents an improved speech dereveraberation method based on the microphone array. The proposed method first calculates the minimum phase of the noisy speech in cepstrum domain, and carries out the beamforming and lowpass filtering. Then the method makes use of the fact that human ear is not sensitive to the phase in- formation, and adopts the phase of the noisy speech to synthesize enhanced speech signals. So the proposed method can reduce the com- putation load greatly. The simulation results show that the proposed method can improve the speech quality effectively in the reverberation room.