摘要
针对说话人识别的GMM模型中参数估计问题,提出了改进的EM算法,利用算法通过对说话人识别系统的实验数据的计算,验证了该算法能提高说话人识别系统的识别率。
The modified EM algorithm for estimating parameters in the GMM model on speakers＇ recognition was proposed. With the modified EM algorithm and the testing data from speakers＇ recognition ,this algorithm was verified to have improved the recognition rate in the system concerned.
出处
《辽宁工业大学学报（自然科学版）》
2009年第3期204-207,共4页
Journal of Liaoning University of Technology(Natural Science Edition)
基金
辽宁省教育厅资助项目(20060409)
关键词
高斯混和模型
说话人识别
EM算法
Gauss Mixture Model （GMM）
speaker recognition
EM algorithm
