Spatial features selection for unsupervised speaker segmentation and clustering

详细信息    查看全文

• 作者：Beatriz Martí ; nez-Gonzá ; lez^a ; ^{beatrizmartinez@die.upm.es} ; José ; M. Pardo^a ; ^{pardo@die.upm.es} ; Juliá ; n D. Echeverry-Correa^b ; ^{jde@utp.edu.co} ; Ruben San-Segundo^a ; ^{lapiz@die.upm.es}
• 关键词：Speaker diarization ; Speaker segmentation ; Feature selection ; Speaker localization ; Speaker clustering
• 刊名：Expert Systems with Applications
• 出版年：2017
• 出版时间：1 May 2017
• 年：2017
• 卷：73
• 期：Complete
• 页码：27-42
• 全文大小：1002 K
• 卷排序：73

文摘

Spatial features selection methods for speaker segmentation are proposed. The system relies on cross correlation measures to perform the selection. The system is compared favourably with other methods such as PCA or the dynamic margin method also developed. The feature selection algorithm reduces both the error rate and the computational cost.