摘要
语音识别是当今信息领域标志性技术,是最具发展前景的综合性学科之一。该文用语音识别系统模式匹配原理,对比分析了模板匹配算法DTW与人工神经网络中的ANN算法在语音识别据识中的应用;提出了一种基于单个神经元的拒识算法,实验结果表明该算法能达到较高的拒识率。
Speech recognition is a sign of information technology and is one of the most promising comprehensive disciplines.This paper briefly describes the speech recognition system and its pattern matching theory.Comparative analysis is made of the template matching algorithms DTW and artificial neural network algorithm ANN for rejection of speech recognition system.A single neuron rejection algorithm is proposed.The results shows that the algorithms achieve a high rejection rate.
出处
《广西师范学院学报（自然科学版）》
2012年第4期88-91,共4页
Journal of Guangxi Teachers Education University(Natural Science Edition)
基金
北京市科委重点项目:物联网关键技术-接入系统研究(No.025000546611003)
211高层次人才培养项目(No.025000543111502)
关键词
语音识别
DTW
人工神经网络
拒识
speech recognition
DTW
Artificial Neural Networks
rejection
