| 基于区别特征检测的汉语韵母分类 |
| |
| 引用本文： | 王孟杰,孟子厚.基于区别特征检测的汉语韵母分类[J].电声技术,2011,35(9):38-41. |
| |
| 作者姓名： | 王孟杰 孟子厚 |
| |
| 作者单位： | 中国传媒大学传播声学研究所,北京,100024 |
| |
| 摘 要： | 从张家騄提出的韵母区别特征树出发,研究区别特征的声学参数.将韵母分为无介音和介音韵母两部分,使用支持向量机检测韵母的区别特征.在此基础上,通过区别特征树上的二元搜索过程实现韵母的分类.每个节点对应于一个区别特征,经过特定的搜索路径,韵母就被唯一确定下来.使用上述方法,大部分韵母的识别率在90％以上.
|
| 关 键 词： | 语音评测 韵母 介音 区别特征 支持向量机 |
Classification of Mandarin Finals Based on Distinctive Features Detection |
| |
| WANG Mengjie,MENG Zihou.Classification of Mandarin Finals Based on Distinctive Features Detection[J].Audio Engineering,2011,35(9):38-41. |
| |
| Authors: | WANG Mengjie MENG Zihou |
| |
| Institution: | (Communication Acoustics Laboratory,Communication University of China,Beijing 100024,China) |
| |
| Abstract: | Acoustic parameters of distinctive feature are studied based on distinctive features framework proposed by Zhang Jialu.Finals are divided into beginning with and without glides.In both cases,two-class classifiers are constructed by support vector machine to check the distinctive features.On this basis,finals are classified through binary search in distinctive features trees.Each node corresponds to a kind of distinctive feature and finals are identified according to a specific search process.The accuracy of most mandarin finals is above 90%. |
| |
| Keywords: | speech evaluation finals glides distinctive features support vector machine |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
