一种汉语语音基音轨迹的平滑算法

基音是指人们发浊音时声带振动所引起的周期性,而声带振动频率的倒数就是基音周期.语音信号处理中精确提取特定语音的基音周期有重要的意义,基音周期的检测无论在语音信号的合成、编码还是识别方面,都起着非常重要的作用.该文依据数字语音信号的特点,设计了一种用于语音处理的层叠滤波算法,合理选择结构元素构筑了一组用于汉语语音基音轨迹平滑的层叠滤波器,仿真实验结果表明,这种方法对于随机误判点与传统的基于中值与线性平滑的组合平滑相比效果更佳,同时也证明了层叠滤波器用于语音信号研究的可行性.

An algorithm for pitch contrail smoothing of Chinese speech

Pitch is the periodicity of vocal cord's vibrations when we pronounce the sonants.The reciprocal of frequency of vocal cord's vibration is the pitch period.It is essential to get accurate pitch period in speech signal processing.The detection of pitch period plays an important role in speech synthesis,speech coding and speech recognition.In this paper,based on the characteristics of speech signals,a new stack filtering algorithm was proposed,in which a new stack filter with proper structural elements was designed for pitch contrail smoothing of speech signal.Simulation results show that this new approach is better in dealing with the stochastic misjudged values than the combined filter based on median filter and linear filter.Furthermore,it indicates that the stack filter is suitable for speech signal research.