首页 | 本学科首页   官方微博 | 高级检索  
     检索      

自回归语音信号模型参数的韧性估计算法及在共振峰检测中的应用
引用本文:江太辉,张有为.自回归语音信号模型参数的韧性估计算法及在共振峰检测中的应用[J].五邑大学学报(自然科学版),1997(1).
作者姓名:江太辉  张有为
作者单位:五邑大学信息科学研究室
基金项目:广东省高教厅重点科研项目资助
摘    要:使用牛顿算法的韧性(Robust)M估计方法来估计自回归(AR)语音信号模型的线性预测参数,该方法考虑了语音信号模型浊音激励源的非高斯特性,能有效地抑制异常值(outlier)的影响。使用所提出的韧性线预测算法来检测共振峰,获得了比使用常规线性预测方法更准确的结果

关 键 词:线性预测  韧性M估计  自回归(AR)信号模型

A Robust Estimate Algorithm For The Linear Prediction Parameters Of Autoregressive(AR)Speech Signal Model and its Application in Detecting The Formants
Jiang Taihui Zhang Youwei.A Robust Estimate Algorithm For The Linear Prediction Parameters Of Autoregressive(AR)Speech Signal Model and its Application in Detecting The Formants[J].Journal of Wuyi University(Natural Science Edition),1997(1).
Authors:Jiang Taihui Zhang Youwei
Institution:Jiang Taihui Zhang Youwei
Abstract:In general, the linear prediction (LP) parameters are estimated by either the autocorrelation method or covariance method. Robust M-estimate procedure on the Newton-type algorithm is used to estimate LP parameters of autoregressive (AR) speech signal model in this paper. The procedure takes into account the non-Gaussian nature of excitations for voiced speech, and it can effectively cut off influence of the outliers in the exciations for voiced speech. The experimental results have shown that the results obtained by the robust LP algorithm in detecting formant center frequencies are more accurate than those by convertional LP method.
Keywords:Linear prediction  Robust M-estimate  Autoregressive (AR) model
本文献已被 CNKI 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号