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

一种基于改进的矢量量化算法的说话人识别方法
引用本文:张云雁,张建军,赵风光,吴龙梅.一种基于改进的矢量量化算法的说话人识别方法[J].上海大学学报(自然科学版),2005,11(4):368-371.
作者姓名:张云雁  张建军  赵风光  吴龙梅
作者单位:上海大学,理学院,上海,200444
基金项目:上海高校发展基金资助项目(214348)
摘    要:该文针对LBG算法可能有空胞腔产生及有些码字利用率低的问题,提出了一种改进的矢量量化算法,并将其应用到与文本无关的说话人识别研究,得到了一种新的说话人识别方法.实验表明,这种方法对说话人的识别性能好于基于LBG算法的说话人识别方法.

关 键 词:矢量量化  LBG算法  说话人识别  LPC倒谱系数
文章编号:1007-2861(2005)04-0368-04
修稿时间:2004年6月21日

A Speaker Recognition Method Based on Modified VQ Algorithm
ZHANG Yun-yan,ZHANG Jian-jun,ZHAO Feng-guang,WU Long-mei.A Speaker Recognition Method Based on Modified VQ Algorithm[J].Journal of Shanghai University(Natural Science),2005,11(4):368-371.
Authors:ZHANG Yun-yan  ZHANG Jian-jun  ZHAO Feng-guang  WU Long-mei
Abstract:The traditional LBG algorithm is an important VQ method, but it has two disadvantages. First, empty cells may be produced, and secondly, some code vectors may be used only by very few training vectors in the LBG algorithm. In this paper, a modified LBG algorithm is proposed, and the design of a text-independent speaker recognition system based on the modified VQ algorithm is presented. Experiments show that the system has better performance than that based on the traditional LBG algorithm.
Keywords:vector quantization  LBG algorithm  speaker recognition  LPC cepstral coefficient  
本文献已被 CNKI 万方数据 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