Publication Details

Hybrid word-subword decoding for spoken term detection

SZŐKE, I.; FAPŠO, M.; BURGET, L.; ČERNOCKÝ, J. Hybrid word-subword decoding for spoken term detection. Proc. SSCS 2008: Speech search workshop at SIGIR. Singapore: Association for Computing Machinery, 2008. p. 1-4. ISBN: 978-90-365-2697-5.
Czech title
Hybridní slovní a podslovní dekóodování pro detekci klíčových frází v řeči
Type
conference paper
Language
English
Authors
Szőke Igor, Ing., Ph.D. (DCGM)
Fapšo Michal, Ing., Ph.D.
Burget Lukáš, doc. Ing., Ph.D. (DCGM)
Černocký Jan, prof. Dr. Ing. (DCGM)
URL
Keywords

spoken term detection

Abstract

The paper is hybrid word-subword decoding for spoken term detection

Annotation

This paper deals with a hybrid word-subword recognition system for spoken term detection. The decoding is driven by a hybrid recognition network and the decoder directly produces hybrid word-subword lattices. One phone and two multigram models were tested to represent sub-word units. The systems were evaluated in terms of spoken term detection accuracy and the size of index. We concluded that the best subword model for hybrid word-subword recognition is the multigram model trained on the word recognizer vocabulary. We achieved an improvement in word recognition accuracy, and in spoken term detection accuracy when in-vocabulary and out-of-vocabulary terms are searched separately. Spoken term detection accuracy with the full (in-vocabulary and out-of-vocabulary) term set was slightly worse but the required index size was significantly reduced.

Published
2008
Pages
1–4
Proceedings
Proc. SSCS 2008: Speech search workshop at SIGIR
ISBN
978-90-365-2697-5
Publisher
Association for Computing Machinery
Place
Singapore
BibTeX 
@inproceedings{BUT32318,
  author="Igor {Szőke} and Michal {Fapšo} and Lukáš {Burget} and Jan {Černocký}",
  title="Hybrid word-subword decoding for spoken term detection",
  booktitle="Proc. SSCS 2008: Speech search workshop at SIGIR",
  year="2008",
  pages="1--4",
  publisher="Association for Computing Machinery",
  address="Singapore",
  isbn="978-90-365-2697-5",
  url="http://www.fit.vutbr.cz/research/groups/speech/publi/2008/szoke_sigir2008.pdf"
}
Back to top