This paper proposes a classification scheme that combines statistical models and support vector machines. It exploits the fact that GMM and SVM classifiers with roughly the same level of performance produce uncorrelated errors. We describe a novel scheme which employs an SVM classifier as an ``advisor'' to the GMM classifier in uncertain cases. The utility of the combined generative/discriminative approach is demonstrated on standard text-independent speaker verification and speaker identification tasks in matched and mismatched training and test conditions. Results indicate significant improvements in performance without much computational overhead.
Cite as: Fine, S., Navratil, J., Gopinath, R.A. (2001) Enhancing GMM scores using SVM "hints". Proc. 7th European Conference on Speech Communication and Technology (Eurospeech 2001), 1757-1760, doi: 10.21437/Eurospeech.2001-411
@inproceedings{fine01_eurospeech,
author={Shai Fine and Jiri Navratil and Ramesh A. Gopinath},
title={{Enhancing GMM scores using SVM "hints"}},
year=2001,
booktitle={Proc. 7th European Conference on Speech Communication and Technology (Eurospeech 2001)},
pages={1757--1760},
doi={10.21437/Eurospeech.2001-411}
}