摘要
给出了加窗语音的新倒谱模型 ,提出了基于语音LPC预测残差的倒谱基音周期检测算法 (LP -CEP法 ) ,该算法在倒谱计算中采取高频分量置零的新措施 ,仿真研究结果表明 ,该算法相对于传统倒谱基音检测算法 (CEP法 ) ,无论是对不同的干净语音还是对不同信噪比的含噪语音 。
A new cepstrum model of windowing speech ,a new speech pitch detection algorithm(LP-CEP),which uses the cepstrum of the LPC predictive residual, is proposed. The method of removing high frequency components is also used for the cepstrum calculation in this algorithm. Simulations showed that the pitch extracting results of the LP-CEP are superior to that of the conventional cepstrum algorithm both for different speeches and in different SNR of a speech.
出处
《南阳师范学院学报》
CAS
2004年第9期26-29,共4页
Journal of Nanyang Normal University
关键词
基音周期
倒谱
线性预测编码
预测残差
pitch period
cepstrum
liner predictive coding (LPC)
predictive residual
