Confidence measure based unsupervised target model adaptation for speaker verification
Résumé
This paper proposes a new method for updating online the client models of a speaker recognition system using the test data. This problem is called unsupervised adaptation. The main idea of the proposed approach is to adapt the client model using the complete set of data gathered from the successive test, without deciding if the test data belongs to the client or to an impostor. The adaptation process includes a weighting scheme of the test data, based on the a posteriori probability that a test belongs to the targeted client model. The proposed approach is evaluated within the framework of the NIST 2005 and 2006 Speaker Recognition Evaluations. The links between the adaptation method and channel mismatch factors is also explored, using both Feature Mapping and Latent Factor Analysis (LFA) methods. The proposed unsupervised adaptation outperforms the baseline system, with a relative DCF improvement of 27% (37% for EER). When the LFA channel compensation technique is used, the proposed approach achieves a reduction in DCF of 20% (12.5% for EER).