Journal of Shanghai University(Natural Science Edition)

• Articles • Previous Articles     Next Articles

An Improved BIC Algorithm for Speaker Change Detection

YU Xiao-qing, TAN Hai-ying   

  1. School of Communication and Information Engineering, Shanghai University, Shanghai 200072, China
  • Received:2007-03-14 Revised:1900-01-01 Online:2007-08-20 Published:2007-08-20
  • Contact: YU Xiao-qing

Abstract: To detect the voice change of speakers, an improved BIC algorithm is proposed in this paper. The algorithm chooses the segmentation point, and then analyzes the trend to locate the voice changing point of speakers. Precision of the algorithm is 90.2% and the recall rate is 95.8%. Compared with the traditional BIC algorithm, this new algorithm has lower computational complexity, with precision and recall rate improved by 62.5% and 16% respectively.

Key words: changing point, precision, recall rate
,
segmentation, BIC detection

CLC Number: