Spatial features selection for unsupervised speaker segmentation and clustering

作者：

Highlights：

• 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.

摘要

•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.

论文关键词：Speaker diarization,Speaker segmentation,Feature selection,Speaker localization,Speaker clustering

论文评审过程：Received 7 June 2016, Revised 14 November 2016, Accepted 5 December 2016, Available online 8 December 2016, Version of Record 24 December 2016.

论文官网地址：https://doi.org/10.1016/j.eswa.2016.12.005