计算机工程 ›› 2009, Vol. 35 ›› Issue (7): 180-182.doi: 10.3969/j.issn.1000-3428.2009.07.063

• 人工智能及识别技术 • 上一篇    下一篇

VoIP压缩码流说话人识别研究

唐 晖,李弼程,屈 丹,张连海   

  1. (解放军信息工程大学信息工程学院,郑州 450002)
  • 收稿日期:1900-01-01 修回日期:1900-01-01 出版日期:2009-04-05 发布日期:2009-04-05

Research on Speaker Recognition from Compressed VoIP Packet Stream

TANG Hui, LI Bi-cheng, QU Dan, ZHANG Lian-hai   

  1. (Institute of Information Engineering, PLA Information Engineering University, Zhengzhou 450002)
  • Received:1900-01-01 Revised:1900-01-01 Online:2009-04-05 Published:2009-04-05

摘要: 研究基于微聚类算法的VoIP压缩码流说话人识别算法。给出直接从G.729, G.723.1(6.3 Kb/s), G.723.1(5.3 Kb/s)压缩语音的码流中提取识别参数,以微聚类算法作为识别结构的说话人识别算法。实验结果表明,对比在压缩码流中使用同样识别参数的GMM模型,微聚类算法在识别正确率和效率上都有很大的提高。

关键词: 说话人识别, 微聚类, 压缩参数

Abstract: This paper presents compressed speaker recognition approach for VoIP(Voice over IP) which is based on the micro-clustering algorithm. It designs a framework based on the micro-clustering algorithm and performs speaker recognition on the feature vector which is directly extracted from G.729, G723.1(6.3 Kb/s), G723.1(5.3 Kb/s) compressed stream. Experimental result shows that the new method is more accurate and efficient than the widely used Gaussian Mixture Model(GMM) which uses the same feature vector.

Key words: speaker recognition, micro-clustering, compressed parameter

中图分类号: