INTERSPEECH 2004 - ICSLP
8th International Conference on Spoken Language Processing

Jeju Island, Korea
October 4-8, 2004

A New Score Normalization Method for Speaker Verification with Virtual Impostor Model

Woo-Yong Choi (1), Jung Gon Kim (2), Hyung Soon Kim (2), Sung Bum Pan (1)

(1) Electronics and Telecommunications Research Institute, Korea
(2) Pusan National University, Korea, Korea

We present a score normalization method, which is suitable for embedded speaker verification system. Proposed score normalization method does not need physical or actual impostor model but utilizes the client model's mean and covariance vector to imitate an impostor model. Therefore, no additional memory for impostor model is required. Furthermore, most parts of score evaluation process for impostor model are identical with that of client model. As a result, computational burden for virtual cohort model is trivial. when comparing with un-normalized likelihood score based speaker verification system, average error rate reduction of the proposed system is higher than 6.7%.

Full Paper

Bibliographic reference.  Choi, Woo-Yong / Kim, Jung Gon / Kim, Hyung Soon / Pan, Sung Bum (2004): "A new score normalization method for speaker verification with virtual impostor model", In INTERSPEECH-2004, 2373-2376.