基于小波变换的传声器阵列语音增强方法

A method for microphone array speech enhancement based on wavelet transform

针对现有的基于传声器阵列语音增强算法的局限性,并考虑到人耳听觉感知模型,提出一种将延迟-求和波束形成技术和小波变换技术相结合进行语音增强的方法.该方法首先利用延迟-求和波束形成技术将阵列中各个传声器接收到的信号进行时间延迟补偿,并对各通道信号相加-平均,消除一部分不相干或弱相干噪声;然后再利用小波变换技术进一步去除噪声.计算机模拟结果表明,该方法具有良好的消噪能力.

A method using delaysum beamforming technique and wavelet transform to eliminate noise based on auditory model is proposed. The signals are received by a microphone array with 4 microphones and the speech enhancement algorithm is implemented in two steps. First, the delaysum beamforming technique is used to eliminate incoherent noise and convert four signals to one. Second, the wavelet transform is used to enhance the speech signal ulteriorly. The computer experiments show that the proposed method has good ability for speech enhancement.