| HTK参数对普米语孤立词识别率的影响 |
| |
| 作者单位： | ;1.云南民族大学数学与计算机科学学院;2.云南民族大学云南省高校物联网应用技术重点实验室 |
| |
| 摘 要： | 普米语是一种无文字的少数民族语言,目前已处于濒危状态,建立语音语料库及开展语音识别研究是保护和传承普米语的重要手段.基于HTK的语音识别中,参数的选取对不同语言的识别率有很大的影响.针对MFCC维数、HMM状态数及GMM个数这3个参数对普米语的识别率进行研究,结果表明:普米语语音识别的最佳MFCC维数为13维,最佳HMM状态数为8个,最佳GMM个数为3个.
|
| 关 键 词： | HTK 普米语 孤立词 语音识别 MFCC HMM GMM |
Influence of HTK parameters on the recognition rate of the isolated words of the Pumi language |
| |
| Institution: | ,School of Mathematics and Computer Science,Yunnan Minzu University,Key Laboratory of IOT Application Technology of Universities in Yunnan Province,Yunnan Minzu University |
| |
| Abstract: | The Pumi language is a spoken language of the Pumi minority group,and is now an endangered language. The establishment of its speech corpora and the relevant research on its speech recognition are two important means to protect and inherit the Pumi language. The selection of parameters based on HTK has much impact on the recognition rate of different languages. This paper focuses on the recognition rate of the Pumi language based on the following three parameters: the MFCC dimension,the number of states in Hidden- Markov- Models( HMM) and the GMM- based number. The results have showed that when the MFCC dimension is 13,the number of HMM state is 8 and the GMM- based number is 3,the recognition rate of the Pumi language is the best. |
| |
| Keywords: | HTK Pumi language isolated word speech recognition MFCC HMM GMM |
| 本文献已被 CNKI 等数据库收录！ |
|
