基于传声器阵列超增益波束形成的语音增强——时域方法

Speech Enhancement Based on Super Gain Beamforming of Microphone Array-Time Domain Approach

给出了传声器阵列宽带超增益波束形成的时域实现方法。在该方法中,传声器阵列各个阵元输出的语音信号先经过数字延迟线,实现整数倍采样间隔的时延补偿,然后由FIR数字滤波器来模拟超增益波束形成所需的不同频率上不同的幅度和相位加权,最后再把FIR数字滤波器的输出相加即得到时域宽带波束输出。仿真了间距为0.05m的5元均匀线性传声器阵列接收到的端射方向带噪线性调频信号和语音信号,并进行时域宽带超增益处理。仿真结果表明,超增益处理比常规处理的阵增益高8.2dB左右,且具有良好的语音增强效果。

A speech enhancement approach based on super gain beamforming of microphone array in time domain is proposed in this paper. The first step of this approach is to delay the outputs of each element in the microphone array by digital delay lines to accomplish the integer part of the time delay. Then, finite impulse response （FIR） digital filters are used to implement the varying amplitude and phase weiglating corresponding to the requirements of super gain beamforming for each element. Finally, the outputs of the digital filters are summed up to obtain the time domain beam output. Computer simulations were carried out for a five element superdirective microphone array with 0.05 m inter-element spacing. Simulation results show that the array gain of super gain processing is 8.2 dB higher than conventional beamforming.