This paper addresses the topic of online unsupervised speaker segmentation in a complex audio environment as it is present in the Broadcast News databases. A new two stage speaker change detection algorithm is proposed, which combines the Bayesian Information Criterion with an ABLS-SCD statistical framework where adapted Gaussian mixture models are used to achieve higher accuracy. To enhance the performance of the proposed method a sub-window dependent threshold selection strategy for the ABLS-SCD is introduced. Also an additional window selection strategy for the proposed method is presented. Experimental design and test evaluation were carried out on the Slovenian BNSI Broadcast News database.
Cite as: Grašič, M., Kos, M., Žgank, A., Kačič, Z. (2008) Two step speaker segmentation method using Bayesian information criterion and adapted Gaussian mixtures models. Proc. Interspeech 2008, 2514-2517, doi: 10.21437/Interspeech.2008-623
@inproceedings{grasic08_interspeech,
author={Matej Grašič and Marko Kos and Andrej Žgank and Zdravko Kačič},
title={{Two step speaker segmentation method using Bayesian information criterion and adapted Gaussian mixtures models}},
year=2008,
booktitle={Proc. Interspeech 2008},
pages={2514--2517},
doi={10.21437/Interspeech.2008-623}
}