Conference proceeding
Maximum Likelihood Linear Dimension Reduction of Heteroscedastic Feature for Robust Speaker Recognition
2015 12TH IEEE INTERNATIONAL CONFERENCE ON ADVANCED VIDEO AND SIGNAL BASED SURVEILLANCE (AVSS), pp 1-5
01 Aug 2015
Abstract
This paper analyzes heteroscedasticity in i-vector for robust forensics and surveillance speaker recognition system. Linear DiscriminantA nalysis (LDA), a widely-used linear dimension reduction technique, assumes that classes are homoscedastic within a same covariance. In this paper it is assumed that general speech utterances contain both homoscedastic and heteroscedastic elements. We show the validity of this assumption by employing several analyses and also demonstrate that dimension reduction using principal components is feasible. To effectively handle the presence of heteroscedastic and homoscedastic elements, we propose a fusion approach of applying both LDA and Heteroscedastic-LDA (HLDA). The experiments are conducted to show its effectiveness and compare to other methods using the telephone database of National Institute of Standards and Technology (NIST) Speaker Recognition Evaluation (SRE) 2010 extended.
Metrics
1 Record Views
2 citations in Scopus
Details
- Title
- Maximum Likelihood Linear Dimension Reduction of Heteroscedastic Feature for Robust Speaker Recognition
- Creators
- Suwon Shon - Korea UniversitySeongkyu Mun - Korea UniversityDavid K. Han - Office of Naval ResearchHanseok Ko - Korea University
- Publication Details
- 2015 12TH IEEE INTERNATIONAL CONFERENCE ON ADVANCED VIDEO AND SIGNAL BASED SURVEILLANCE (AVSS), pp 1-5
- Publisher
- IEEE
- Number of pages
- 5
- Resource Type
- Conference proceeding
- Language
- English
- Academic Unit
- Electrical and Computer Engineering
- Web of Science ID
- WOS:000380619700064
- Scopus ID
- 2-s2.0-84958625592
- Other Identifier
- 991021931087704721
InCites Highlights
Data related to this publication, from InCites Benchmarking & Analytics tool:
- Collaboration types
- Domestic collaboration
- International collaboration
- Web of Science research areas
- Imaging Science & Photographic Technology
- Transportation Science & Technology