相干滤波与广义旁瓣相消器结合的小阵列语音增强算法

Speech Enhancement Method Based on Combination of Coherence Filter and Generalized Sidelobe Canceler for Small Microphone Arrays

相干滤波器与广义旁瓣相消器(GSC)是常用的阵列语音增强算法,然而,应用于小阵列中却存在消噪能力不足的问题。针对上述问题,本文提出一种相干滤波与广义旁瓣相消器结合的小阵列语音增强算法。首先,利用广义旁瓣相消器对带噪语音进行初步增强。然后,通过改进的最小搜索算法估计出信号里残余噪声的功率谱密度,从而获得相干滤波器的传递函数。最后,利用相干滤波器对带噪语音进行再次增强。仿真实验表明:在多种不同的噪声环境下该算法具有较好的噪声抑制能力。

Coherence filter and generalized sidelobe canceler( GSC) are common methods of array speech enhancement,however they are lack of the denoising ability in a small array. This paper puts forward a small array speech enhancemet method which is combined with coherence filter and GSC. At first,noisy speech was initially enhanced by applying generalized sidelobe canceler. Afterwards,noise power spectral density was estimated via improved minimum tracking approach. So transfer function of coherence filter was obtained.Finally,noisy speech was enhanced once again with the help of coherence filter. Simulation experimentindicates that under different noisy circumstances,the calculation method has a relative efficient effect on noisy suppression.