机构地区: 西安电子科技大学
出 处: 《现代电子技术》 2006年第2期43-45,48,共4页
摘 要: 在介绍Hilbert Huang变换理论的基础上,提出了一种基于HHT变换的语音去噪算法。首先对带噪语音信号做EMD分解,得到各阶IMF分量,然后对高频的IMF分量用小波域去噪中的阈值方法进行处理,然后把经过阈值处理的高频IMF分量和低频的IMF进行叠加,得到重构后的信号,即去噪信号。仿真实验表明基于Hilbert Huang变换的去噪结果优于小波软、硬阈值法的去噪结果,显示了Hilbert Huang变换在处理非平稳信号中的优越性。 Hilbert- Huang Transform (HHT) is a new and effective method of analyzing nonlinear and non- stationary time series. In this paper,a denoising method of speech signal based on HHT is presented. First,speech signal polluted by white noise is decomposed into several Intrinsic Mode Functions (IMF) based on Hilbert- Huang transform. Then,the intrinsic mode functions of high frequency is preprocessed using threshold method,and we add these IMFs with IMFs of low frequency to achieve denoising signal. Simulation experiments show that the results based on HHT are better than that of soft or hard threshold method.