Detail publikace

Czech Speech Recognizer for Multiple Environments

GLEMBEK, O.; KARAFIÁT, M.; BURGET, L.; ČERNOCKÝ, J. Czech Speech Recognizer for Multiple Environments. Radioeletronika 2006. Bratislava: 2006. p. 1-4.
Název česky
Rozpoznávač češtiny pro různá prostředí
Typ
článek ve sborníku konference
Jazyk
anglicky
Autoři
Glembek Ondřej, Ing., Ph.D.
Karafiát Martin, Ing., Ph.D. (UPGM)
Burget Lukáš, doc. Ing., Ph.D. (UPGM)
Černocký Jan, prof. Dr. Ing. (UPGM)
URL
Klíčová slova

speech, recognition, automatic, artiffical inteligence, training, czech, database, acoustic, modelling, modeling, language

Abstrakt

This paper presents our work on building a largevocabulary continuousspeech recognition (LVCSR) system for Czech, capable of op eration inmultiple environments. SpeeCon and Temic speech databases were used todefine a data-set for training acoustic models, attention was paid tounification of these two resources. The test set was also defined usingthese corp ora with careful choice of segments not overlapping with thetraining data. The system was completed by a language model trained onCzech National corpus. The recognition was performed using DUCoder anLVCSR stack decoder. Experimental results on the LVCSR task give areference score of the system for future improvements.

Rok
2006
Strany
1–4
Sborník
Radioeletronika 2006
Konference
16. mezinárodní Česko-Slovenská vědecká konference Radioelektronika 200ž, Bratislava, SK
Místo
Bratislava
BibTeX 
@inproceedings{BUT22376,
  author="Ondřej {Glembek} and Martin {Karafiát} and Lukáš {Burget} and Jan {Černocký}",
  title="Czech Speech Recognizer for Multiple Environments",
  booktitle="Radioeletronika 2006",
  year="2006",
  pages="1--4",
  address="Bratislava",
  url="http://www.fit.vutbr.cz/~glembek/papers/radioeletronika_2006.pdf"
}
Nahoru