Detail výsledku

Neural Network Bottleneck Features for Language Identification

MATĚJKA, P.; ZHANG, L.; NG, T.; MALLIDI, S.; GLEMBEK, O.; MA, J.; ZHANG, B. Neural Network Bottleneck Features for Language Identification. In Proceedings of Odyssey 2014. Proceedings of Odyssey: The Speaker and Language Recognition Workshop Odyssey 2014, Joensuu, Finland. Joensuu: International Speech Communication Association, 2014. no. 6, p. 299-304. ISSN: 2312-2846.
Typ
článek ve sborníku konference
Jazyk
anglicky
Autoři
Matějka Pavel, Ing., Ph.D., UPGM (FIT)
Zhang Le
Ng Tim
Mallidi Sri Harish, FIT (FIT)
Glembek Ondřej, Ing., Ph.D., UPGM (FIT)
Ma Jeff
Zhang Bing
Abstrakt

We have presented the bottleneck features in the context of Languageidentification. It combines benefits of both phonotacticand acoustic system. Usually, the phonotactic system is favorablefor the long duration files, while acoustic for the short ones.This approach takes the advantage of both. In addition, we canalso use modeling of context dependent phonemes in bottleneckfeatures. This brings very nice improvement over the contextindependent phonemes.

Klíčová slova

language identification, noisy speech, robustfeature extraction

URL
Rok
2014
Strany
299–304
Časopis
Proceedings of Odyssey: The Speaker and Language Recognition Workshop Odyssey 2014, Joensuu, Finland, roč. 2014, č. 6, ISSN 2312-2846
Sborník
Proceedings of Odyssey 2014
Konference
Odyssey 2014: The Speaker and Language Recognition Workshop
Vydavatel
International Speech Communication Association
Místo
Joensuu
EID Scopus
2-s2.0-85073163761
BibTeX 
@inproceedings{BUT111630,
  author="Pavel {Matějka} and Le {Zhang} and Tim {Ng} and Sri Harish {Mallidi} and Ondřej {Glembek} and Jeff {Ma} and Bing {Zhang}",
  title="Neural Network Bottleneck Features for Language Identification",
  booktitle="Proceedings of Odyssey 2014",
  year="2014",
  journal="Proceedings of Odyssey: The Speaker and Language Recognition Workshop Odyssey 2014, Joensuu, Finland",
  volume="2014",
  number="6",
  pages="299--304",
  publisher="International Speech Communication Association",
  address="Joensuu",
  issn="2312-2846",
  url="http://www.fit.vutbr.cz/research/groups/speech/publi/2014/matejka_odyssey2014_299-304-35.pdf"
}
Projekty
DARPA Robustní automatický přepis řeči (RATS) - RATS Patrol I, BBN, zahájení: 2010-09-23, ukončení: 2014-06-30, ukončen
Výzkumné skupiny
Výzkumná skupina dolování dat z řeči BUT Speech@FIT (VZ SPEECH)
Pracoviště
Ústav počítačové grafiky a multimédií (UPGM)
Nahoru