Result Details

Generating Exact Lattices in The WFST Framework

POVEY, D.; HANNEMANN, M.; BOULIANNE, G.; BURGET, L.; GHOSHAL, A.; JANDA, M.; KARAFIÁT, M.; KOMBRINK, S.; MOTLÍČEK, P.; QIAN, Y.; RIEDHAMMER, K.; VESELÝ, K.; VU, N. Generating Exact Lattices in The WFST Framework. Proceedings of 2012 IEEE International Conference on Acoustics, Speech and Signal Processing. Kyoto: IEEE Signal Processing Society, 2012. p. 4213-4216. ISBN: 978-1-4673-0044-5.
Type
conference paper
Language
English
Authors
Povey Daniel
Hannemann Mirko, Ph.D., DCGM (FIT)
Boulianne Gilles
Burget Lukáš, doc. Ing., Ph.D., DCGM (FIT)
Ghoshal Arnab
Janda Miloš, Ing., DCGM (FIT)
Karafiát Martin, Ing., Ph.D., DCGM (FIT)
Kombrink Stefan, Dipl.-Linguist., DCGM (FIT)
Motlíček Petr, doc. Ing., Ph.D., DCGM (FIT)
Qian Yanmin
Riedhammer Korbinian
Veselý Karel, Ing., Ph.D., DCGM (FIT)
Vu Ngoc Thang
Abstract

This article describes a lattice generation method, which is most directly applicable when using WFST decoders.

Keywords

Speech Recognition, Lattice Generation

URL
Annotation

We describe a lattice generation method that is exact, i.e. it satisfies all the natural properties we would want from a lattice of alternative transcriptions of an utterance. This method does not introduce substantial overhead above one-best decoding. Our method is most directly applicable when using WFST decoders where the WFST is "fully expanded", i.e. where the arcs correspond to HMM transitions. It outputs lattices that include state-level alignments as well as word labels. The general idea is to create a state-level lattice during decoding, and to do a special form of determinization that retains only the best-scoring path for each word sequence. This special determinization algorithm is a solution to the following problem: Given a WFST A, compute a WFST B that, for each input-symbolsequence of A, contains just the lowest-cost path through A. Our solution always works if the projection of A on its input labels is determinizable, e.g. if A is acyclic.

Published
2012
Pages
4213–4216
Proceedings
Proceedings of 2012 IEEE International Conference on Acoustics, Speech and Signal Processing
Conference
The 37th International Conference on Acoustics, Speech, and Signal Processing
ISBN
978-1-4673-0044-5
Publisher
IEEE Signal Processing Society
Place
Kyoto
DOI
10.1109/ICASSP.2012.6288848
BibTeX 
@inproceedings{BUT91476,
  author="Daniel {Povey} and Mirko {Hannemann} and Gilles {Boulianne} and Lukáš {Burget} and Arnab {Ghoshal} and Miloš {Janda} and Martin {Karafiát} and Stefan {Kombrink} and Petr {Motlíček} and Yanmin {Qian} and Korbinian {Riedhammer} and Karel {Veselý} and Ngoc Thang {Vu}",
  title="Generating Exact Lattices in The WFST Framework",
  booktitle="Proceedings of 2012 IEEE International Conference on Acoustics, Speech and Signal Processing",
  year="2012",
  pages="4213--4216",
  publisher="IEEE Signal Processing Society",
  address="Kyoto",
  doi="10.1109/ICASSP.2012.6288848",
  isbn="978-1-4673-0044-5",
  url="http://www.fit.vutbr.cz/research/groups/speech/publi/2012/povey_icassp2012_0004213.pdf"
}
Projects
Security-Oriented Research in Information Technology, MŠMT, Institucionální prostředky SR ČR (např. VZ, VC), MSM0021630528, start: 2007-01-01, end: 2013-12-31, running
Speech Recognition under Real-World Conditions, GACR, Standardní projekty, GA102/08/0707, start: 2008-01-01, end: 2011-12-31, completed
Technologies of speech processing for efficient human-machine communication, TAČR, Program aplikovaného výzkumu a experimentálního vývoje ALFA, TA01011328, start: 2011-01-01, end: 2014-12-31, completed
Research groups
Speech Data Mining Research Group BUT Speech@FIT (RG SPEECH)
Departments
Department of Computer Graphics and Multimedia (DCGM)
Back to top