应用分段辨认序列频度信息的说话人确认方法  

A Text-independent Speaker Verification Method Based on the Information of Candidate Sequences of Speech Segment-identification

作　　者：翁武斌 方棣棠[1] 

机构地区：[1]清华大学计算机科学与技术系

出　　处：《中文信息学报》2000年第5期31-36,共6页Journal of Chinese Information Processing

摘　　要：本文提出了一种基于语音分段辨认序列信息的与文本无关的说话人确认方法 ,并且着重分析了其中关键因素的变化 ,包括聚类数、阈值以及判定准则的变化 ,对确认效果的影响。通过实验证明了分段辨认序列频度信息是一种非常有效的说话人确认信息 ,对于确认结果起到很好的辅助作用。同时也指出了新方法的不足和今后的改进方向。In this paper,a text independent speaker verification method based on the information of candidate sequences of speech segment identification is described.And the influences caused by the changes of the key factors in the method such as the number of the cluster centers,the threshold and the judge rules are analyzed emphatically.Through experiments it has been proved that the in forma tion of candidate sequences of speech segment identification works very well in speaker verification and it is an effective auxiliary method in improving the performance of speaker verification system.Finally,the deficiency of the new method is pointed out and future work is depicted.

关 键 词：说话人确认 语音识别 分段辨认序列信息 

分 类 号：TN912.34[电子电信—通信与信息系统]