Publication Details

BUT- PT System Description for NIST LRE 2017

MATĚJKA, P.; PLCHOT, O.; NOVOTNÝ, O.; CUMANI, S.; LOZANO DÍEZ, A.; SLAVÍČEK, J.; DIEZ SÁNCHEZ, M.; GRÉZL, F.; GLEMBEK, O.; KAMSALI VEERA, M.; SILNOVA, A.; BURGET, L.; ONDEL YANG, L.; KESIRAJU, S.; ROHDIN, J. BUT- PT System Description for NIST LRE 2017. Proceedings of NIST Language Recognition Workshop 2017. Orlando, Florida: National Institute of Standards and Technology, 2017. p. 1-6.
Type
conference paper
Language
English
Authors
Matějka Pavel, Ing., Ph.D. (DCGM)
Plchot Oldřich, Ing., Ph.D. (DCGM)
Novotný Ondřej, Ing., Ph.D.
Cumani Sandro, Ph.D.
LOZANO DÍEZ, A.
SLAVÍČEK, J.
DIEZ SÁNCHEZ, M.
Grézl František, Ing., Ph.D. (DCGM)
Glembek Ondřej, Ing., Ph.D.
KAMSALI VEERA, M.
Silnova Anna, M.Sc., Ph.D. (DCGM)
Burget Lukáš, doc. Ing., Ph.D. (DCGM)
ONDEL YANG, L.
Kesiraju Santosh, Ph.D. (DCGM)
Rohdin Johan Andréas, M.Sc., Ph.D. (DCGM)
URL
Keywords

speech recognition, language recognition

Abstract

This article is about the BUT - PT System Description for the NIST LRE 2017 evaluation. We have built over 30 systems for this evaluation with the main focus to build a  single best system. We experimented with denoising NN, automatic discovery units, different flavors of phonotactic systems, different backends, different sizes of i-vector systems, different BN features, NN embeddings and frame level language classifiers. The evaluation plan stated "Teams are encouraged to report whether and how having access to the development set helped improve the performance". The development data helped mainly in the final classifier and also helped in the decision process which techniques to use and which to fuse because our test set consisted of this data.

Annotation

Our submission is a collaborative effort of BUT, Politecnico di Torino, Universidad Autonoma de Madrid and Phonexia. The main body of work was conducted during end of September and beginning of October 2017 when the whole team met in Brno and all members were closely working together with common datasets. All of our individual systems rely on the bottleneck features[1, 2] (BNF) as frontends. Most of our systems are still based on i-vectors and subsequent generative classifier. We also complement the classical i-vector based systems with a system based on embeddings obtained from discriminatively trained end-toend LRE system. Finally, the primary submission is a fusion of four systems where we utilize two different BNF extractors, non-linear processing of i-vectors and embeddings obtained from the discriminative system.

Published
2017
Pages
1–6
Proceedings
Proceedings of NIST Language Recognition Workshop 2017
Publisher
National Institute of Standards and Technology
Place
Orlando, Florida
BibTeX 
@inproceedings{BUT168463,
  author="MATĚJKA, P. and PLCHOT, O. and NOVOTNÝ, O. and CUMANI, S. and LOZANO DÍEZ, A. and SLAVÍČEK, J. and DIEZ SÁNCHEZ, M. and GRÉZL, F. and GLEMBEK, O. and KAMSALI VEERA, M. and SILNOVA, A. and BURGET, L. and ONDEL YANG, L. and KESIRAJU, S. and ROHDIN, J.",
  title="BUT- PT System Description for NIST LRE 2017",
  booktitle="Proceedings of NIST Language Recognition Workshop 2017",
  year="2017",
  pages="1--6",
  publisher="National Institute of Standards and Technology",
  address="Orlando, Florida",
  url="https://www.fit.vut.cz/research/publication/11655/"
}
Files
Back to top