On the enhancement of speaker identification accuracy using weighted bilateral scoring

Abstract

???This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder." ???Copyright IEEE. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.??? DOI: 10.1109/CCST.2008.4751310This paper presents investigations into an effective bilateral scoring method in open-set speaker identification. The approach is based on the fact that two different speakers usually are not reciprocal. A difficulty in deploying bilateral scoring is that test utterances are normally much shorter than training utterances. To tackle this problem, the proposed approach provides the final identification score based on a weighted combination of independently normalised forward and reverse scores. Based on the experimental results obtained using clean and telephone quality speech, it is shown that the proposed approach is more effective than the conventional scoring methods in open-set speaker identification

Open-set speaker identification with diverse-duration speech data

“…Weighted bilateral scoring (WBS) 6 provides an extension of this traditional approach. In the context of this study, its potential benefit is related to a scenario in the GMM-UBM paradigm, in which the training utterance (utterance x) from a particular speaker is too short, whereas the testing utterance (utterance y) from the same speaker is considerably longer.…”

“…However, bilateral scoring involves combining the above forward score with a reverse score obtained by matching utterance x against the richer model obtained through the use of utterance y. It should be further emphasised that fusing the reverse score with the traditional GMM-UBM forward score can be specifically beneficial for the real-world applications, which are likely to involve reference and test speech data of varied lengths [6][7][8] .…”

A Fishervoice based feature fusion method for short utterance speaker recognition

“…Some methods try to improve the performance by selecting segments with higher discriminability on speaker characteristics [8]. The weighted bilateral scoring method is used to improve the performance of speaker recognition in the scoring domain [9]. Obviously, most of these foresaid approaches can obtain benefits by utilizing models or data containing higher information for speaker recognition and all of them achieve performance improvement with utterance length of 5~10 seconds.…”