论文部分内容阅读
文中在小波包变换和传统阈值法的基础上,提出了一种基于小波包变换的多尺度多阈值语音信号去噪净化方法。采用小波包分解,克服了传统的正交小波变换的缺陷。采用多尺度多阈值方法,通过改进噪声方差估计方法,在去噪的同时,进一步提高信噪比。仿真实验结果表明,本方法能有效去除信号中的噪声和较好保留语音细节,达到更佳的语音净化效果。
Based on wavelet packet transform and traditional thresholding method, a novel denoising and purification method for multi-scale and multi-threshold speech signals based on wavelet packet transform is proposed. The wavelet packet decomposition overcomes the defects of traditional orthogonal wavelet transform. Using multi-scale and multi-thresholding method, by improving the noise variance estimation method, the signal to noise ratio is further improved while denoising. Simulation results show that this method can effectively remove the noise in the signal and retain the voice details better, to achieve better voice purification effect.