| 语音识别中隐马尔可夫模型状态数的研究 |
| |
| 引用本文： | 张焱,张杰.语音识别中隐马尔可夫模型状态数的研究[J].南京理工大学学报(自然科学版),1998,22(3):208-211,215. |
| |
| 作者姓名： | 张焱 张杰 |
| |
| 作者单位： | 南京理工大学信息学院 |
| |
| 基金项目： | 江苏省自然科学基金,南京理工大学科研发展基金 |
| |
| 摘 要： | 该文从信息论的观点出发，对语音信号的隐马尔可夫模型（ＨＭＭ）的状态数进行研究，建立了ＨＭＭ的状态数研究的简化模型，指出ＨＭＭ的信息熵是由语音信号的固有熵和附加熵组成。随状态数增加，信息熵趋向固有熵。最后，在综合考虑信息熵和运算量两方面因素情况下，得出了状态数宜在６￣８之间的结论。
|
| 关 键 词： | 语音识别 信息论 熵 隐马尔可夫模型 状态数 |
Study on State's Number of HMM of Speech Recognition |
| |
| Zhang Yan,Zhang Jie,Huang Zhitong.Study on State's Number of HMM of Speech Recognition[J].Journal of Nanjing University of Science and Technology(Nature Science),1998,22(3):208-211,215. |
| |
| Authors: | Zhang Yan Zhang Jie Huang Zhitong |
| |
| Abstract: | This paper investigates state's number of HMM (Hidden Markov Model) of speech signal on the view of information theory, constructs a simplified mathematical model for studying the number of state of HMM. The results show that the information entropy of speech signal is constituted of intrinsic entropy and additional entropy. With the number of state increasing, the information entropy tends towards its intrinsic entropy. The conclusion is obtained that the number of state of HMM is properly between 6 and 8 by considering information entropy and calculation. |
| |
| Keywords: | speech recognition information theory entropy hidden Markov model |
| 本文献已被 CNKI 维普 等数据库收录！ |
|
