基于信号相位差和后置滤波的语音增强方法

 在复杂的声学环境中,通常噪声场特性和混响强度是未知的,这样就对麦克风阵列语音增强算法的性能提出了较高的要求.本文提出一种基于带噪语音信号相位差和后置滤波的语音增强方法.首先,将麦克风阵列接收信号分帧,利用相邻两个麦克风之间每帧带噪语音信号的相位差,构成该帧改变频率点幅度谱值的比例系数,对该帧带噪语音信号进行掩蔽增强处理,得到预处理信号;然后利用固定波束形成、独立分量分析算法和后置滤波技术对预处理信号进一步处理,从而有效地抑制了噪声.计算机仿真实验结果表明,在存在一定混响的多种噪声场中,该方法均具有较好的噪声抑制能力.

A Speech Enhancement Method Based on Phrase-Error and Post-Filtering

In complicated acoustic environment where the characteristics of noise field and reverberation time are unknown,the performance of the speech enhancement methods with microphone array face a great challenge.A speech enhancement method based on phrase-errors of the noisy speech signals and post-filter is proposed.First,in each frame of the microphone array signals,by using the phrase-errors of the adjacent noisy speech signals,we can construct scaling coefficients to preprocess the noisy speeches,and then th...