改进的最小均方误差语音增强算法的研究

A study of an improved minimum mean-square error speech enhancement algorithm

针对传统最小均方误差谱幅度估计(MMSE-STSA,minimum mean-square error-short time spectral am-plitude)语音增强算法无法有效的跟踪非平稳噪声变化的问题,对一种改进的MMSE-STSA语音增强算法进行了研究和仿真。该算法对背景噪声的估计利用加权噪声估计方法:采用一个非线性函数根据带噪语音信噪比(SNR,signal-to-noise ratio)的变化计算得到相应的加权因子并作用于带噪语音信号,对加权的带噪语音求平均得到估计的背景噪声。算法中的谱增益修正,还可以抑制低信噪比时的残留噪声以及避免对带噪语音的过抵消。实验结果表明,该方法能很好的跟踪非平稳噪声的变化,不仅在增强性能上有很好的效果,同时降低了语音的失真。

To achieve good tracking capability without overestimation of various non-stationary noise sources, an improved MMSE-STSA speech enhancement algorithm is proposed in this paper. The algorithm employs weighted noise estimation： the noisy speech is weighted by a weighting factor, which is calculated in accordance with the estimated SNR through a nonlinear function, and the estimated noise is obtained as an average of the weighted noisy speech. The spectrum gain modification in this algorithm can further suppress the residual noise for low SNRs and avoid the excessive suppression. The results indicate that, under the non-stationary noisy environment, the proposed algorithm can not only get a good performance in enhancement, but also reduce the speech distortion.