基于子带TF-GSC麦克风阵列语音增强

Microphone Array Speech Enhancement Based on Subband Transfer Function Generalized Sidelobe Canceller

为了加快基于传递函数广义旁瓣相消器的麦克风阵列语音增强系统的收敛速度,将其自适应模块的输入信号分解到子带以进行处理,并将多通道维纳滤波器引入传递函数广义旁瓣相消器的非自适应支路,以便更有效地抑制非相干噪声。实际测试结果表明,相对于基于全带广义旁瓣相消器的麦克风阵列语音增强系统,采用该子带传递函数广义旁瓣相消器结构的语音增强系统具有更高的输出信噪比。

In order to accelerate the convergence of the microphone array speech enhancement system based on Transfer Function Generalized Sidelobe Canceller（ TF - GSC） , the input signals of the adaptive block are decomposed into theirs subbands to perform the adaptive multi- channel noise canceling in each subband, And the integration of multi - channel wiener filter into the fixed branch of TF - GSC can suppress the uncorrelated noise efficiently. Experiments indicate that the microphone array speech enhancement system based on this suhhand TF - GSC has better output signal - to - noise ratio than its counterpart.