基于语音分组的基音周期检测方法

针对语音信号的特点，清音和浊音，清音没有明显的周期性，它类似于高斯白噪声，浊音具有周期性的特征，并且浊音的基音周期只有准周期的特点，另外语音信号是非线性时变的信号，只能采用短时平均方法估计基音周期。针对这些特点，本文对处理的语音信号按照频率进行分组，分组后再进行基音周期的提取，减少了处理的复杂度和增加了准确率。

The pitch of the voice packet detection method based on

In this paper, according to the features of speech signals, speech and voiced, unvoiced speech has no obvious periodicity, which is similar to the Gauss white noise, character voiced with pe- riodic, pitch and voicing only quasi-periodic characteristics, the speech signal is nonlinear and time-va- rying, only using the short time average method of pitch estimation. According to these characteris- tics, the speech signal processing are grouped according to the frequency, packet after extraction of pitch period, reduce the processing complexity and increases the accuracy rate.