In this paper we explore the use of Variational Bayesian (VB) learning in unsupervised speaker clustering. VB learning is a relatively new learning technique that has the capacity of doing at the same time parameter learning and model selection. We tested this approach on the NIST 1996 HUB-4 evaluation test for speaker clustering when the speaker number is a priori known and when it has to be estimated. VB shows a higher accuracy in terms of average cluster purity and average speaker purity compared to the Maximum Likelihood solution.
Cite as: Valente, F., Wellekens, C. (2004) Variational Bayesian speaker clustering. Proc. The Speaker and Language Recognition Workshop (Odyssey 2004), 207-214
@inproceedings{valente04_odyssey,
author={Fabio Valente and Christian Wellekens},
title={{Variational Bayesian speaker clustering}},
year=2004,
booktitle={Proc. The Speaker and Language Recognition Workshop (Odyssey 2004)},
pages={207--214}
}