基于小波变换的静音与语音分割新算法

New noisy speech segmentation algorithm based on wavelet transform

含噪语音信号的静音与语音分割 ,即端点检测问题是语音识别至关重要的一步 .为了提高语音分割对环境的适应性 ,提出了一种利用小波变换分割含噪语音信号中静音与语音的新算法 .该算法首先将语音信号进行小波变换 ,利用小波系数去噪 ,然后选择小波部分子带跟踪信号的能量变化以分割语音与静音 .

Segmentation of noisy speech signals, namely detection of speech signals endpoint, is the key for speech recognition. A new noisy segmentation algorithm based on the wavelet transform is proposed to improve the adaptability of speech segmentation to environments. Wavelet transform is first employed to denoise, and the cross correlation between the wavelet coefficients in two sub-bands is then used to extract pure speech duration. Simulations were made at different signal-to-noise ratios and the results show that this method is efficient to segment noisy speech even at a low signal-to-noise ratio.