基于掩蔽效应的维纳滤波器语音增强及DSP实现

针对单通道音增强,常用的谱减法及其改进型算法等语音增强方法不能很好地消除残留音乐噪声以及不能兼顾噪声抑制和语音失真的问题,提出用维纳滤波法与掩蔽效应相结合的算法。介绍了DSP的实现方法。结果表明,该算法相对于传统谱减法和改进谱减法语音增强效果显著、失真小、音乐噪声几乎没有,并且本算法计算量小,很容易在基于DSP的语音增强算法中实现。

Speech enhancement and DSP implementation of Wiener filter based on masking effect

For the single channel speech enhancement,the residual music noise cannot be reduced completely,and the speech distortion cannot be given attention when suppressing noise in the traditional spectral subtraction method.A speech enhancement method of which combined of Wiener filter and masking effect is proposed.At last,the implementation method based on DSP is in Introduced.The simulation results prove that,the new method has the better effects of speech enhancement,and the distortion is lower,the music noises disappear almost.It can easily transplant to the DSP because of the small amount of calculation.