Result Details

Discriminatively Trained Probabilistic Linear Discriminant Analysis for Speaker Verification

BURGET, L.; PLCHOT, O.; CUMANI, S.; GLEMBEK, O.; MATĚJKA, P.; BRÜMMER, N. Discriminatively Trained Probabilistic Linear Discriminant Analysis for Speaker Verification. In Proceedings of the 2011 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2011. Praha: IEEE Signal Processing Society, 2011. p. 4832-4835. ISBN: 978-1-4577-0537-3.
Type
conference paper
Language
English
Authors
Burget Lukáš, doc. Ing., Ph.D., DCGM (FIT)
Plchot Oldřich, Ing., Ph.D., DCGM (FIT)
Cumani Sandro, Ph.D.
Glembek Ondřej, Ing., Ph.D., DCGM (FIT)
Matějka Pavel, Ing., Ph.D., DCGM (FIT)
Brümmer Niko
Abstract

This paper is on Discriminatively Trained Probabilistic Linear Discriminant Analysis (PLDA) for Speaker Verification.

Keywords

Speaker verification, Discriminative training, Probabilistic Linear Discriminant Analysis

URL
Annotation

Recently, i-vector extraction and Probabilistic Linear Discriminant Analysis (PLDA) have proven to provide state-of-the-art speaker verification performance. In this paper, the speaker verification score for a pair of i-vectors representing a trial is computed with a functional form derived from the successful PLDA generative model. In our case, however, parameters of this function are estimated based on a discriminative training criterion. We propose to use the objective function to directly address the task in speaker verification: discrimination between same-speaker and different-speaker trials. Compared with a baseline which uses a generatively trained PLDA model, discriminative training provides up to 40% relative improvement on the NIST SRE 2010 evaluation task.

Published
2011
Pages
4832–4835
Proceedings
Proceedings of the 2011 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2011
Conference
International Conference on Acoustics, Speech and Signal Processing 2011
ISBN
978-1-4577-0537-3
Publisher
IEEE Signal Processing Society
Place
Praha
DOI
10.1109/ICASSP.2011.5947437
EID Scopus
2-s2.0-80051611495
BibTeX 
@inproceedings{BUT76384,
  author="Lukáš {Burget} and Oldřich {Plchot} and Sandro {Cumani} and Ondřej {Glembek} and Pavel {Matějka} and Niko {Brümmer}",
  title="Discriminatively Trained Probabilistic Linear Discriminant Analysis for Speaker Verification",
  booktitle="Proceedings of the 2011 IEEE International Conference on Acoustics, Speech, and Signal Processing, ICASSP 2011",
  year="2011",
  pages="4832--4835",
  publisher="IEEE Signal Processing Society",
  address="Praha",
  doi="10.1109/ICASSP.2011.5947437",
  isbn="978-1-4577-0537-3",
  url="https://www.fit.vut.cz/research/publication/9653/"
}
Projects
IARPA Biometrics Exploitation Science and Technology (BEST) - Promoting Robustness in Speaker Modeling (PRISM), IARPA, start: 2009-12-07, end: 2011-12-30, completed
Security-Oriented Research in Information Technology, MŠMT, Institucionální prostředky SR ČR (např. VZ, VC), MSM0021630528, start: 2007-01-01, end: 2013-12-31, running
Speech Recognition under Real-World Conditions, GACR, Standardní projekty, GA102/08/0707, start: 2008-01-01, end: 2011-12-31, completed
Research groups
Speech Data Mining Research Group BUT Speech@FIT (RG SPEECH)
Departments
Department of Computer Graphics and Multimedia (DCGM)
Back to top