Improvements of search error risk minimization in viterbi beam search for speech recognition

Takaaki Hori, Shinji Watanabe, Atsushi Nakamura

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

This paper describes improvements in a search error risk minimization approach to fast beam search for speech recognition. In our previous work, we proposed this approach to reduce search errors by optimizing the pruning criterion. While conventional methods use heuristic criteria to prune hypotheses, our proposed method employs a pruning function that makes a more precise decision using rich features extracted from each hypothesis. The parameters of the function can be estimated to minimize a loss function based on the search error risk. In this paper, we improve this method by introducing a modified loss function, arc-averaged risk, which potentially has a higher correlation with actual error rate than the original one. We also investigate various combinations of features. Experimental results show that further search error reduction over the original method is obtained in a 100K-word vocabulary lecture speech transcription task.

Original languageEnglish
Title of host publicationProceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010
Pages1962-1965
Number of pages4
Publication statusPublished - 2010
Externally publishedYes
Event11th Annual Conference of the International Speech Communication Association: Spoken Language Processing for All, INTERSPEECH 2010 - Makuhari, Chiba
Duration: 2010 Sep 262010 Sep 30

Other

Other11th Annual Conference of the International Speech Communication Association: Spoken Language Processing for All, INTERSPEECH 2010
CityMakuhari, Chiba
Period10/9/2610/9/30

Fingerprint

Vocabulary
Recognition (Psychology)
Speech Recognition
Heuristics
Conventional
Transcription

Keywords

  • Beam search
  • Pruning
  • Search error
  • Speech recognition
  • WFST

ASJC Scopus subject areas

  • Language and Linguistics
  • Speech and Hearing

Cite this

Hori, T., Watanabe, S., & Nakamura, A. (2010). Improvements of search error risk minimization in viterbi beam search for speech recognition. In Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010 (pp. 1962-1965)

Improvements of search error risk minimization in viterbi beam search for speech recognition. / Hori, Takaaki; Watanabe, Shinji; Nakamura, Atsushi.

Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010. 2010. p. 1962-1965.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Hori, T, Watanabe, S & Nakamura, A 2010, Improvements of search error risk minimization in viterbi beam search for speech recognition. in Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010. pp. 1962-1965, 11th Annual Conference of the International Speech Communication Association: Spoken Language Processing for All, INTERSPEECH 2010, Makuhari, Chiba, 10/9/26.
Hori T, Watanabe S, Nakamura A. Improvements of search error risk minimization in viterbi beam search for speech recognition. In Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010. 2010. p. 1962-1965
Hori, Takaaki ; Watanabe, Shinji ; Nakamura, Atsushi. / Improvements of search error risk minimization in viterbi beam search for speech recognition. Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010. 2010. pp. 1962-1965
@inproceedings{6627452ef22b4163b33e4c4ba5572e5f,
title = "Improvements of search error risk minimization in viterbi beam search for speech recognition",
abstract = "This paper describes improvements in a search error risk minimization approach to fast beam search for speech recognition. In our previous work, we proposed this approach to reduce search errors by optimizing the pruning criterion. While conventional methods use heuristic criteria to prune hypotheses, our proposed method employs a pruning function that makes a more precise decision using rich features extracted from each hypothesis. The parameters of the function can be estimated to minimize a loss function based on the search error risk. In this paper, we improve this method by introducing a modified loss function, arc-averaged risk, which potentially has a higher correlation with actual error rate than the original one. We also investigate various combinations of features. Experimental results show that further search error reduction over the original method is obtained in a 100K-word vocabulary lecture speech transcription task.",
keywords = "Beam search, Pruning, Search error, Speech recognition, WFST",
author = "Takaaki Hori and Shinji Watanabe and Atsushi Nakamura",
year = "2010",
language = "English",
pages = "1962--1965",
booktitle = "Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010",

}

TY - GEN

T1 - Improvements of search error risk minimization in viterbi beam search for speech recognition

AU - Hori, Takaaki

AU - Watanabe, Shinji

AU - Nakamura, Atsushi

PY - 2010

Y1 - 2010

N2 - This paper describes improvements in a search error risk minimization approach to fast beam search for speech recognition. In our previous work, we proposed this approach to reduce search errors by optimizing the pruning criterion. While conventional methods use heuristic criteria to prune hypotheses, our proposed method employs a pruning function that makes a more precise decision using rich features extracted from each hypothesis. The parameters of the function can be estimated to minimize a loss function based on the search error risk. In this paper, we improve this method by introducing a modified loss function, arc-averaged risk, which potentially has a higher correlation with actual error rate than the original one. We also investigate various combinations of features. Experimental results show that further search error reduction over the original method is obtained in a 100K-word vocabulary lecture speech transcription task.

AB - This paper describes improvements in a search error risk minimization approach to fast beam search for speech recognition. In our previous work, we proposed this approach to reduce search errors by optimizing the pruning criterion. While conventional methods use heuristic criteria to prune hypotheses, our proposed method employs a pruning function that makes a more precise decision using rich features extracted from each hypothesis. The parameters of the function can be estimated to minimize a loss function based on the search error risk. In this paper, we improve this method by introducing a modified loss function, arc-averaged risk, which potentially has a higher correlation with actual error rate than the original one. We also investigate various combinations of features. Experimental results show that further search error reduction over the original method is obtained in a 100K-word vocabulary lecture speech transcription task.

KW - Beam search

KW - Pruning

KW - Search error

KW - Speech recognition

KW - WFST

UR - http://www.scopus.com/inward/record.url?scp=79959812951&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79959812951&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:79959812951

SP - 1962

EP - 1965

BT - Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010

ER -