基于DTW和LVQ网络混合模型的语音识别方法  

Speech Recognition Method Based on Hybrid Model of Dynamic Time Warping and Learning Vector Quantization

作　　者：林遂芳[1] 张海英[1] 潘永湘[1] 

机构地区：[1]西安理工大学自动化与信息工程学院,陕西西安710048

出　　处：《系统仿真学报》2005年第8期1959-1961,1965,共4页Journal of System Simulation

摘　　要：提出一种基于动态时间规整(DTW)和学习矢量量化(LVQ)神经网络的语音识别方法。该方法用动态时间规整算法先对语音信号进行时间规整,然后通过学习矢量量化神经网络进行语音的分类识别。首先介绍利用动态时间规整和学习矢量量化进行语音识别的基本方法,然后给出DTW/LVQ混合模型的系统结构和学习算法,最后给出三种语音识别算法的实验结果。大量实验表明,混合模型的识别率,皆明显高于单一的动态时间规整和学习矢量量化的识别率。A novel method for speech recognition tasks in Chinese language is presented. The structure of a speaker-independent hybrid model for isolated word recognition, based on Learning Vector Quantization （LVQ） technique combined with the Dynamic Time Warping （DTW） algorithm is described. The Dynamic Time Warping algorithm is used as the front-end of Learning Vector Quantization network, warping the epoch of the input utterances. The task of classification and recognition is completed by the Learning Vector Quantization network, which is modified in the learning algorithm. The experimental results demonstrate that the hybrid model leads to higher recognition rates than the classic technologies.

关 键 词：语音识别 动态时间规整 学习矢量量化 混合模型 

分 类 号：TN912.34[电子电信—通信与信息系统]