A New Re-estimation Algorithm of HMM’s Transition Probability

LI Jian; WANG Zuo-ying

您当前的位置：

首页 >

文章列表页 >

A New Re-estimation Algorithm of HMM’s Transition Probability

更新时间：2025-07-16

- A New Re-estimation Algorithm of HMM’s Transition Probability
- Acta Electronica Sinica Vol. 29, Issue S1, Pages: 1833-1835(2001)
- 作者机构：
  
  清华大学电子工程系,北京,100084
- 作者简介：
- 基金信息：
- DOI：
  CLC： TP39l
- Published：2001
- 稿件说明：
移动端阅览
LI Jian, WANG Zuo-ying. A New Re-estimation Algorithm of HMM’s Transition Probability[J]. Acta Electronica Sinica, 2001, 29(S1): 1833-1835.
DOI：

LI Jian, WANG Zuo-ying. A New Re-estimation Algorithm of HMM’s Transition Probability[J]. Acta Electronica Sinica, 2001, 29(S1): 1833-1835. DOI：

摘要

将隐含马尔可夫模型(HMM:Hidden Markov Model)引入到语音识别中来是一个巨大的贡献.但是在经典的HMM中关于状态转移概率a

(#em/em#

)与自转移概率a

的独立性假设

导致了这个模型的不协调性.事实上

段长分布概率与状态转移概率并非相互独立的

由其中的一个就可以唯一的确定另外一个.本文从段长分布概率出发说明了以上关于转移概率独立性假设的不合理性

并得到了转移概率新的重估算法.这个新算法比经典HMM的Baum-Welch迭代算法重估转移概率效果更好

前者比后者相对误识率下降了大约5%.

Abstract

It is a great contribution that introduces hidden Markov model(HMM)in speech recognition.But the assumption of the traditional HMM that state transition probabilities a

(#em/em#

)are independent of self-trainsiton probabilty a

leads to disharmomy to the model.In fact

duration distribution probability and transition probability are dependent and one can be determined by another. From the viewpoint of duration distribution probabilty

this paper shows that the aussumption of independence of transition probability is unreasonable.And a new re-estimation algorithm of transition probability is deduced.The new algorithm performs better than the Baum-Welch algorithm.The average error reduction is about 5%.

关键词

Keywords

references

Views

1303

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Convergence Rate Estimation of Elitist Clonal Selection Algorithm

Related Author

HONG Lu

GONG Cheng-long

WANG Jing-zhuo

JI Zhi-cheng

Related Institution

Department of Electronic Engineering, Huaihai Institute of Technology

Department of Internet of Things Engineering, Jiangnan University

Department of Electronic Engineering Huaihai Institute of Technology Lianyungang Jiangsu China

Department of Internet of Things Engineering Jiangnan University Wuxi Jiangsu China

⁰