EMD算法和Hilbert算法混合的语音信号降噪和提取

De-noise and Extraction of Speech Signal Based on Hybrid of EMD Algorithm and Hilbert Algorithm

针对语音信号降噪和提取算法存在降噪效果不佳等问题,已经无法满足现在社会的需求。本文提出一种基于改进EMD算法和Hilbert算法混合的语音信号降噪和提取方法,首先针对EMD算法在语音信号降噪可能破坏噪声特性的问题,利用全局阈值选择去噪的方法对其进行改进,然后引入Hil-bert算法将其与EMD算法融合,对语音信号进行降噪和提取。利用仿真实验可以发现:将EMD算法和Hilbert算法进行混合优化,和传统的小波变换语音信号降噪算法,不容易出现差错并且具有更好的降噪性能。

As poor performance of de-noise in algorithm of speech signal de-noises and extraction, italready cannot satisfy the demand of the society now. A method of speech signal de- noises andextraction is proposed based on hybrid improved EMD algorithm and Hilbert algorithm. First of all, asthe EMD algorithm could undermine the noise characteristics in speech signal de-noising, so improve itby using global threshold de-noising methods, and then introduce the Hilbert algorithm fusion with theEMD algorithm, de-noise and extract the speech signals. Get the results from simulation experiments：compared with the speech signal de- noising algorithm of traditional wavelet transformation, hybridoptimization algorithm of the EMD and Hilbert algorithm is not easy to error and has better performanceof de-noise.