白噪声下基于各向异性滤波的语音增强算法研究

Study of Algorithm for Speech Enhancement Based on Anisotropic Diffusion in White Noise

通过介绍各向异性滤波的模型,然后用语音信号的幅度去代替图像的梯度,进而利用各向异性滤波在图像处理领域中的滤波原理对语音信号进行处理,分析处理结果,并且将该方法的增强效果与其他三种经典的语音增强算法做了比较.仿真结果表明:在白噪声为背景噪声的情况下,各向异性滤波方法具有较好的滤波性能,并且与其他三种经典的语音增强算法相比,其算法复杂度低,计算量小,运行时间短.

The anisotropic diffusion model was introduced,and the speech signal amplitude was used to replace the image gradient. Then,the speech signal was processed by the filtering theory of the anisotropic diffusion in the image processing field,and the processing results were analyzed. The comparison between the enhanced effect of this methods and the other three classic speech enhancement algorithm was made. The simulation results show that the anisotropic diffusion has the better filtering performance in the background noise of white noise,and the method has the lower algorithm complexity,the smaller calculation amount,the shorter running time compared with the other three classic speech enhancement algorithms.