首页 | 官方网站   微博 | 高级检索  
     

一种自适应线谱对滤波器的终点误差LMS算法及其学习特性
引用本文:江太辉,张有为.一种自适应线谱对滤波器的终点误差LMS算法及其学习特性[J].信号处理,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
摘    要:本文提出了一种计算线谱对(LSP)系数的新算法.该算法以最小均方(LMS)误差型算法为基础,梯度计算使用系统终点误差。对接收到的语音观测样本在一个时序的基础进行迭代更新计算,直接产生LSP系数,形成一个LMS算法的自适应LSP滤波器。实验结果表明,该算法与其它自适应滤器的LMS算法比较,具有更高的收敛率和较低的失调。

关 键 词:自适应格型滤波器  LSP滤波器  线性预测  LMS算法

An End-Point Error LMS Algorithm of Adaptive Line Spectral Pair Filter and Its Learning Characteristics
Jiang Taihui, Zhang Youwei.An End-Point Error LMS Algorithm of Adaptive Line Spectral Pair Filter and Its Learning Characteristics[J].Signal Processing,1997(1).
Authors:Jiang Taihui  Zhang Youwei
Abstract:In this paper, a new algorthm used to calculate line spectral pair (LSP) coefficients is presented . The algorithm is based on the least-mean -square (LMS)type a lgorithm , and the end -point error of system is used for gradient estimation . A updating calculation is carried out for received speech samples on a sequential basis to generate directly LSP coefficients , and a adaptive LSP filter on the basis of LMS algorithmis is formed . Experimental results show that the algorithm has higher convergence rate and low misadjustment as compared with that LMS algorithm of other daptive filters.
Keywords:Adaptive lattice filter  LSP filter  Linear pretiction  LMS algorithm  
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-23

京公网安备 11010802026262号