| 引用本文: | 王成友,梁甸农,唐朝京,等.三种语音特征信息综合方法的理论分析.[J].国防科技大学学报,1999,21(4):48-54.[点击复制] |
| Wang ChengYou,Liang DianNong,Tang Zhaojing,et al.The Theory Study of the Three Kinds of Methods for Combining the Information of Speech Features[J].Journal of National University of Defense Technology,1999,21(4):48-54[点击复制] |
|
| |
|
|
| 本文已被：浏览 5942次 下载 6057次 |
| 三种语音特征信息综合方法的理论分析 |
| 王成友, 梁甸农, 唐朝京, 杨述明 |
|
(国防科技大学 电子技术系 湖南 长沙 410073)
|
| 摘要: |
| 本文是作者在1997年《声学学报》中文版和英文版发表的三篇文章基础上的进一步研究。这些文章没有对提出的三种语音特征综合方法进行理论分析, 本文运用概率论知识, 详细地对前面三篇文章提出的三种语音特征信息综合方法进行理论分析。分析结果表明: 三种语音特征信息综合方法能够比较好地综合语音特征信息, 提高语音识别系统的性能。这一结果与前面三篇文章中实验结果一致。 |
| 关键词: 语言特征信息综合, 顶尖参数法, 全部参数法, 最可靠参数法, 语音识别 |
| DOI： |
| 投稿日期：1999-03-08 |
| 基金项目: |
|
| The Theory Study of the Three Kinds of Methods for Combining the Information of Speech Features |
| Wang ChengYou, Liang DianNong, Tang Zhaojing, Yang shuming |
|
(Department of Electronic Technology, NUDT, Changsha, 410073)
|
| Abstract: |
| This paper is a continuation of three papers that were published in《Chinese Journal of Acoustics》in Chinese version and English version. It discusses three kinds of methods for combining the information of speech features that were presented in detail in the previous three papers in accordance with the theory of probability. The result shows the three methods improve the performance in speech recognition system. That is the same with the experiment result. |
| Keywords: combination of speech features information, the maximum parameter method, all parameter method, the most reliable parameter method, speech recognition |
|
|
|
|
|
