摘要
本文详细介绍了隐 Markov 模型(HMM)的基本概念和计算概率的前、后向算法,并采用 HMM 作了多讲话者(三人)小词汇量汉语单字识别实验。正确识别率为98%。
In this paper the basic concepts of Hidden Markov Model(HMM)and the for- ward and backward algorithm are introduced.Then a word recognition for multi- speaker(three persons)based on HMM is performed.and achieves 98% accuracy
