INTERSPEECH 2011 12th Annual Conference of the International Speech Communication Association Florence, Italy August 27-31. 2011

Speaker Clustering Based on Utterance-Oriented Dirichlet Process Mixture Model

Naohiro Tawara (1), Shinji Watanabe (2), Tetsuji Ogawa (3), Tetsunori Kobayashi (1)

(1) Waseda University, Japan
(2) NTT Corporation, Japan
(3) Waseda Institute for Advanced Study, Japan

This paper provides the analytical solution and algorithm of UO-DPMM based on a non-parametric Bayesian manner, and thus realizes fully Bayesian speaker clustering. We carried out preliminary speaker clustering experiments by using a TIMIT database to compare the proposed method with the conventional Bayesian Information Criterion (BIC) based method, which is an approximate Bayesian approach. The results showed that the proposed method outperformed the conventional one in terms of both computational cost and robustness to changes in tuning parameters.

Full Paper

Bibliographic reference.  Tawara, Naohiro / Watanabe, Shinji / Ogawa, Tetsuji / Kobayashi, Tetsunori (2011): "Speaker clustering based on utterance-oriented dirichlet process mixture model", In INTERSPEECH-2011, 2905-2908.