Back to Main Conference 2008
LREC 2008main

MISTRAL: a Statistical Machine Translation Decoder for Speech Recognition Lattices

Proceedings of the Sixth International Conference on Language Resources and Evaluation (LREC 2008)

DOI:10.63317/5jhson5fhahj

Abstract

This paper presents MISTRAL, an open source statistical machine translation decoder dedicated to spoken language translation. While typical machine translation systems take a written text as input, MISTRAL translates word lattices produced by automatic speech recognition systems. The lattices are translated in two passes using a phrase-based model. Our experiments reveal an improvement in BLEU when translating lattices instead of sentences returned by a speech recognition system.

Details

Paper ID
lrec2008-main-485
Pages
N/A
BibKey
patry-langlais-2008-mistral
Editor
N/A
Publisher
European Language Resources Association (ELRA)
ISSN
2522-2686
ISBN
2-9517408-4-0
Conference
Sixth International Conference on Language Resources and Evaluation
Location
Marrakech, Morocco
Date
28 May 2008 30 May 2008

Authors

  • AP

    Alexandre Patry

  • PL

    Philippe Langlais

Links