Odyssey 2008: The Speaker and Language Recognition Workshop Stellenbosch, South Africa January 21-24, 2008

Feature Vector Classification by Threshold for Speaker identification

Sang-min Yoon, Kyung-mi Park, Jae-Hyun Bae, Yung-hwan Oh

Division of Computer Science Department of EECS, Korea Advanced Institutes of Science and Technology

This paper describes a new feature vector classification method for speaker identification. Purpose of this paper is constructing robust speaker models which only use meaningful feature vectors and discard confusing feature vectors. To construct robust speaker model, proposed method classifies feature vectors using log-likelihood estimation. Experimental results, with various segments ranging from 0.5 to 5s, showed that our method outperforms previous method.

Full Paper

Bibliographic reference.  Yoon, Sang-min / Park, Kyung-mi / Bae, Jae-Hyun / Oh, Yung-hwan (2008): "Feature vector classification by threshold for speaker identification", In Odyssey-2008, paper 006.