基于语音波形外观的基音提取算法

在对人用视觉提取基音周期过程模拟的基础上，提出一种基于语音波形外观形状的时域基音周期提取算法，该算法利用语音波形的一次峰值点和二次峰值点的幅度和位置以及峰到前峰的距离等几种属性，来判断决定基音周期值，具有算法简单、运算量小、能准确定位各基音周期位置的特点。

Pitch EStimation Algorithm Based on Profile of Speech Waveform

On the basis of the simulation of human pitch extraction Process with eyes,a new pitch estimation algorithm hased on the profile of speech waveform is put forward.This algorithm gets pitch periods by applying several attributes, such as the amplitude and position of Peak and quadric Peak of speech waveform and distance from one Peak to the former peak. The algorithm has the following advantages; (l ) it is simple, and with high accuracy (2 ) it doesn' t need too much calculation and (3) itcan locate the pitch peridds. In addition, without window processing, the average effect brought aboutby windowing is absent. When handling speech signals of large difference between two pitch periods,this algorithm has its apparent advantages, so the algorithm is fit for working as a pre--processor of thepitch--synchronous algorithm.