语音信号去噪声处理研究与仿真

Research and Simulation of Noise Cancellation for Speech Signals

语音信号去噪音处理以及浊音、清音的辨别,对智能机器人人机语音交互系统有着重要意义。利用MATLAB软件平台设计了巴特沃斯FIR数字滤波器,进行语音信号去噪音滤波处理。由于原始语音信号的时域波形图与频谱图变化小,分析较难,为便于波形分析,对原始语音信号进行4千赫兹和8千赫兹重采样,通过短时平均过零率与短时自相关函数法对语音信号进行清、浊音的辨别和基音周期提取。仿真结果表明,基于切比雪夫窗函数设计的滤波器对语音信号滤波效果良好。

There are about to deal with noise and distinguish between voiced,unvoiced speech signal,has important significance for the intelligent robot human-computer interaction system.This paper is the use of MATLAB software platform designed by Butterworth FIR digital filter for speech signal to noise filtering,because the original speech signal time-domain waveform and frequency spectrum analysis is difficult to change,in order to facilitate the waveform analysis,the original voice signal of 4 thousand Hz and 8 thousand Hz sampling.Finally,a brief overview is given to the recognition of speech signals and the extraction of pitch period by means of short-time average zero crossing rate and short-time autocorrelation function.The simulation results show that the filter designed by Chebyshev window function is good for speech signal filtering.