| 一种新的声韵分割的时域方法 |
| |
| 引用本文： | 何振武 丁洪. 一种新的声韵分割的时域方法[J]. 数据采集与处理, 1998, 13(2): 154-158 |
| |
| 作者姓名： | 何振武 丁洪 |
| |
| 作者单位： | 上海交通大学仪器工程系 |
| |
| 摘 要： | 针对以前时域特征参数帧平均能量Ｅｎ和帧平均跨零率Ｚｎ在汉语语音声韵分割中的不足，提出了一种新的时域特征参数——保号率ＲＫＳ（Ｋｅｅｐｉｎｇ－ｓｉｇｎ－ｒａｔｉｏ）。实验表明，引入ＲＫＳ参数的汉语语音自动分段系统，不仅从根本上解决了Ｅｎ和Ｚｎ的不足，而且大大提高了自动分段的准确率，降低了对输入语音的依赖性。
|
| 关 键 词： | 语音；时域分析；平均能量；帧平均跨零率；保号率；时域特征参数；自动分段 |
A New Time Domain Method on Separating Consonants and Vowels |
| |
| He Zhenwu Ding Hong Lin Liangming Dept. of Instrument Engineering,ShanghaiJiaotong University Shanghai. A New Time Domain Method on Separating Consonants and Vowels[J]. Journal of Data Acquisition & Processing, 1998, 13(2): 154-158 |
| |
| Authors: | He Zhenwu Ding Hong Lin Liangming Dept. of Instrument Engineering ShanghaiJiaotong University Shanghai |
| |
| Affiliation: | He Zhenwu Ding Hong Lin Liangming Dept. of Instrument Engineering,ShanghaiJiaotong University Shanghai,200030 |
| |
| Abstract: | A new time domain parameter named R KS to resolve the disadvantage of two former time domain parmeters E n and Z n is presented. Experiments show that mardarin speech separating system using R KS parameter absolutely resolves the disadvantage of E n and Z n , raises the accuracy ratio and reduces the dependance on input speech. |
| |
| Keywords: | speech sounds time domain analysis average energy average pass zero ratio keeping sign ratio time domain charactreristic parameter automatic seaparation |
| 本文献已被 CNKI 维普 等数据库收录！ |
