首页 | 本学科首页   官方微博 | 高级检索  
     检索      

基于Mel系数和SVM的语音识别技术研究
引用本文:陈婷,敖茂尧,陈贺璋.基于Mel系数和SVM的语音识别技术研究[J].广西商业高等专科学校学报,2010(5):1-4.
作者姓名:陈婷  敖茂尧  陈贺璋
作者单位:广西职业技术学院,广西南宁530226
摘    要:本文研究了一种运用Mel频标倒谱系数作为特征参数和多分类支持向量机进行语音识别的方法。根据已有的分类方法,尝试将一种基于二叉树的多分类支持向量机用于语音识别。实验表明,基于Mel倒谱和多分类支持向量机的语音识别方法使系统的性能和速度大大提高,具有理想的识别效果和应用价值。

关 键 词:语音识别  Mel倒谱  支持向量机  核函数

Study of Speech Recognition Technology Based on MFCC and SVM
Chen Ting Ao Maoyao Chen Hezhang.Study of Speech Recognition Technology Based on MFCC and SVM[J].Journal of Guangxi Commercial College,2010(5):1-4.
Authors:Chen Ting Ao Maoyao Chen Hezhang
Institution:Chen Ting Ao Maoyao Chen Hezhang (Guangxi Vocational and Technical College Nanning,Guangxi 530226)
Abstract:This paper does research on the method to make speech recognition by using Multi-classification Support Vector Machines and Mel Frequency Cepstrum Coefficient as characteristic parameter. According to the existing classification method,a kind of Multi -classification SVM based on binary tree is used for speech recognition. Experiment results show that the speech recognition method based on MFCC and Multi-classification SVM can greatly improve the system function and the speed of speech recognition,with ideal recognition effect and application value.
Keywords:speech recognition MFCC SVM kernel function
本文献已被 维普 等数据库收录!
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号