| Author: | Cheung, Man-yee Miranda |
| Title: | Decision strategies for robust speaker verification |
| Degree: | M.Sc. |
| Year: | 2001 |
| Subject: | Speech processing systems Automatic speech recognition Hong Kong Polytechnic University -- Dissertations |
| Department: | Multi-disciplinary Studies Department of Electronic Engineering |
| Pages: | vi, 68 leaves : ill. ; 30 cm |
| Language: | English |
| Abstract: | In this dissertation, a new decision strategy based on statistical post-processing of speakers' scores is proposed for speaker verification. Specifically, the dissertation explores the possibility of enhancing the state-of-the-art speaker verification systems by taking into account the discrepancy between speaker's scores and his/her decision threshold. To this end, speaker's scores are ranked according to their proximity to the decision threshold. Non-parametric statistical tests, such as the Wilcoxon signed rank test, are applied to the ranked scores. The Wilcoxon signed rank test is presented in detail and the reasons for choosing it to do the post-processing of speakers' scores are explained. Experimental results, based on 76 speakers of the TIMIT corpus, show that the performance of a vector quantization-based speaker verification system is improved after applying the Wilcoxon test. The method is very general and can be easily extended to other speaker verification systems. |
| Rights: | All rights reserved |
| Access: | restricted access |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| b15995793.pdf | For All Users (off-campus access for PolyU Staff & Students only) | 2.14 MB | Adobe PDF | View/Open |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/396