| 马英,于向飞.一种改进的短时平均幅度差函数算法[J].,2010,29(5):387-390 |
| 一种改进的短时平均幅度差函数算法 |
| An improved absolute magnitude-difference function algorithm |
| |
| 中文摘要: |
| 在语音信号分析中,对于基音周期的提取目前已有较多的分析和处理方法,在现有的短时平均幅度差函数(AMDF)的处理方法中,只需要加、减和取绝对值运算,运算量较之短时自相关函数大大下降。同时,AMDF函数的谷点提取基音周期比自相关函数的峰值更加尖锐,错判率相对较少,稳健性更高。然而,传统的AMDF算法对窗长的要求较为严格,窗长较短就会有较大的误差。本文针对该缺陷做出的改进算法,使之无论窗长多大均会有较为准确的结果,大大拓展了AMDF算法的适用空间。进一步,将其与同态处理结合,会有更好的效果。 |
| 英文摘要: |
| In analysis of speech signal,there are various methods for pitch analysis at present. The method of AMDF,which requiring add subtract and absolute value operations,make self correlation function calculation very efficient.At the same time,pitch cycle,which is extracted by the valley point of AMDF function,is sharper,and is more reliable than the peak of self correlation function.However,The traditional AMDF algorithm has strict for the width of window,otherwise it will have more error.The article is specifically for the improvement of AMDF with two windows sizes.Further,the proposed algorithm will achieve even better results when combined with homomorphic processing. |
| DOI：10.11684/j.issn.1000-310X.2010.05.012 |
| 中文关键词: 短时平均幅度差函数 同态处理 基音周期 MATLAB |
| 英文关键词: AMDF Homomorphic processing Pitch cycle MATLAB |
| 基金项目: |
|
| 摘要点击次数: 2372 |
| 全文下载次数: 963 |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
