摘要
隐马尔可夫模型 (HMM)的基本技术是语音识别中较为成功的算法 .主要是它具有较强的对时间序列结构的建模能力 .本文首先深入浅出地介绍了 HMM的基本技术和一个基于 HMM的孤立词语音识别系统的构成方法 ,其次 ,基于 HMM尚存有一些缺陷 ,造成语音识别能力较弱 ,为此本文又进一步阐述了语音识别应用中的几种改进的 HMM系统及目前的热点方法—— HMM与
The basic technology of HMM is some kind of successful method in dealing with speech recognition. It has a stronger ability to build a model of the time order. This article introduces this basic technology and a speech recognition system on some isolated words. But there are some shortcomings in HMM, which lessens the ability of recognition. On this, this article briefly introduces some reformed HMM speech recognition systems and some popular way the mix net of HMM and ANN.
出处
《工科数学》
2002年第6期16-20,共5页
Journal of Mathematics For Technology
