Publication Details

STBU system for the NIST 2006 speaker recognition evaluation

MATĚJKA, P.; BURGET, L.; SCHWARZ, P.; GLEMBEK, O.; KARAFIÁT, M.; GRÉZL, F.; ČERNOCKÝ, J.; VAN LEEUWEN, D.; BRÜMMER, N.; STRASHEIM, A. STBU system for the NIST 2006 speaker recognition evaluation. Proc. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2007). Honolulu: IEEE Signal Processing Society, 2007. p. 221-224. ISBN: 1-4244-0728-1.
Czech title
STBU systém pro NIST evaluaci rozpoznávání mluvčího 2006
Type
conference paper
Language
English
Authors
Matějka Pavel, Ing., Ph.D. (DCGM)
Burget Lukáš, doc. Ing., Ph.D. (DCGM)
Schwarz Petr, Ing., Ph.D. (DCGM)
Glembek Ondřej, Ing., Ph.D.
Karafiát Martin, Ing., Ph.D. (DCGM)
Grézl František, Ing., Ph.D. (DCGM)
Černocký Jan, prof. Dr. Ing. (DCGM)
van Leeuwen David
Brümmer Niko
Strasheim Albeert
URL
Keywords

Speaker recognition,GMM, SVM, eigenchannel, NAP.

Abstract

This paper describes STBU 2006 speaker recognition system, whichperformed well in the NIST 2006 speaker recognition evaluation. STBU isconsortium of 4 partners: Spescom DataVoice (South Africa), TNO(Netherlands), BUT (Czech Republic) and University of Stellenbosch(South Africa). The primary system is a combination of three main kindsof systems: (1) GMM, with short-time MFCC or PLP features, (2) GMM-SVM,using GMM mean supervectors as input and (3) MLLR-SVM, using MLLRspeaker adaptation coefficients derived from English LVCSR system. Inthis paper, we describe these sub-systems and present results for eachsystem alone and in combination on the NIST Speaker RecognitionEvaluation (SRE) 2006 development and evaluation data sets.

Annotation

This paper describes STBU 2006 speaker recognition system, which performed well in the NIST 2006 speaker recognition evaluation. STBU is consortium of 4 partners: Spescom DataVoice (South Africa), TNO (Netherlands), BUT (Czech Republic) and University of Stellenbosch (South Africa). The primary system is a combination of three main kinds of systems: (1) GMM, with short-time MFCC or PLP features, (2) GMM-SVM, using GMM mean supervectors as input and (3) MLLR-SVM, using MLLR speaker adaptation coefficients derived from English LVCSR system. In this paper, we describe these sub-systems and present results for each system alone and in combination on the NIST Speaker Recognition Evaluation (SRE) 2006 development and evaluation data sets.

Published
2007
Pages
221–224
Proceedings
Proc. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2007)
Conference
32nd IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), Honolulu, US
ISBN
1-4244-0728-1
Publisher
IEEE Signal Processing Society
Place
Honolulu
BibTeX 
@inproceedings{BUT28575,
  author="Pavel {Matějka} and Lukáš {Burget} and Petr {Schwarz} and Ondřej {Glembek} and Martin {Karafiát} and František {Grézl} and Jan {Černocký} and David {van Leeuwen} and Niko {Brümmer} and Albeert {Strasheim}",
  title="STBU system for the NIST 2006 speaker recognition evaluation",
  booktitle="Proc. IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP 2007)",
  year="2007",
  pages="221--224",
  publisher="IEEE Signal Processing Society",
  address="Honolulu",
  isbn="1-4244-0728-1",
  url="http://www.fit.vutbr.cz/research/groups/speech/publi/2007/matejka_stbu_icassp_2007.pdf"
}
Back to top