Using sub-word n-gram models for dealing with OOV in large vocabulary speech recognition for Latvian

Askars Salimbajevs; Jevgenijs Strigins

Using sub-word n-gram models for dealing with OOV in large vocabulary speech recognition for Latvian

Askars Salimbajevs
, Jevgenijs Strigins

Tilde Company

Research output: Chapter in Book/Report/Conference proceeding › Conference paper › Research › peer-review

6 Citations (Scopus)

Abstract

In the Latvian language, one word can have tens or even hundreds of surface forms. This is a serious problem for large vocabulary speech recognition. Inclusion of every form in vocabulary will make it intractable, but, on the other hand, even with a vocabulary of 400K, the out-ofvocabulary (OOV) rate will be very high. In this paper, the authors investigate the possibility of using sub-word vocabularies where words are split into frequent and common parts. The results of our experiment show that this allows to significantly reduce the OOV rate.

Original language	English
Title of host publication	Proceedings of the 20th Nordic Conference of Computational Linguistics, NODALIDA 2015
Publisher	Association for Computational Linguistics (ACL)
Pages	281-285
Number of pages	5
ISBN (Electronic)	9789175190983
ISBN (Print)	9789175190983
Publication status	Published - 2015
Externally published	Yes
Event	20th Nordic Conference of Computational Linguistics, NODALIDA 2015 - Vilnius, Lithuania Duration: 11 May 2015 → 13 May 2015

Publication series

Name	Proceedings of the 20th Nordic Conference of Computational Linguistics, NODALIDA 2015

Conference

Conference	20th Nordic Conference of Computational Linguistics, NODALIDA 2015
Country/Territory	Lithuania
City	Vilnius
Period	11/05/15 → 13/05/15

OECD Field of Science

1.2 Computer and Information Sciences

Cite this

Salimbajevs, A., & Strigins, J. (2015). Using sub-word n-gram models for dealing with OOV in large vocabulary speech recognition for Latvian. In Proceedings of the 20th Nordic Conference of Computational Linguistics, NODALIDA 2015 (pp. 281-285). (Proceedings of the 20th Nordic Conference of Computational Linguistics, NODALIDA 2015). Association for Computational Linguistics (ACL).

Salimbajevs, Askars ; Strigins, Jevgenijs. / Using sub-word n-gram models for dealing with OOV in large vocabulary speech recognition for Latvian. Proceedings of the 20th Nordic Conference of Computational Linguistics, NODALIDA 2015. Association for Computational Linguistics (ACL), 2015. pp. 281-285 (Proceedings of the 20th Nordic Conference of Computational Linguistics, NODALIDA 2015).

@inproceedings{af30fa4b6546452baabc4ae47e6efbae,

title = "Using sub-word n-gram models for dealing with OOV in large vocabulary speech recognition for Latvian",

abstract = "In the Latvian language, one word can have tens or even hundreds of surface forms. This is a serious problem for large vocabulary speech recognition. Inclusion of every form in vocabulary will make it intractable, but, on the other hand, even with a vocabulary of 400K, the out-ofvocabulary (OOV) rate will be very high. In this paper, the authors investigate the possibility of using sub-word vocabularies where words are split into frequent and common parts. The results of our experiment show that this allows to significantly reduce the OOV rate.",

author = "Askars Salimbajevs and Jevgenijs Strigins",

year = "2015",

language = "English",

isbn = "9789175190983",

series = "Proceedings of the 20th Nordic Conference of Computational Linguistics, NODALIDA 2015",

publisher = "Association for Computational Linguistics (ACL)",

pages = "281--285",

booktitle = "Proceedings of the 20th Nordic Conference of Computational Linguistics, NODALIDA 2015",

address = "Australia",

}

Salimbajevs, A & Strigins, J 2015, Using sub-word n-gram models for dealing with OOV in large vocabulary speech recognition for Latvian. in Proceedings of the 20th Nordic Conference of Computational Linguistics, NODALIDA 2015. Proceedings of the 20th Nordic Conference of Computational Linguistics, NODALIDA 2015, Association for Computational Linguistics (ACL), pp. 281-285, 20th Nordic Conference of Computational Linguistics, NODALIDA 2015, Vilnius, Lithuania, 11/05/15.

Using sub-word n-gram models for dealing with OOV in large vocabulary speech recognition for Latvian. / Salimbajevs, Askars; Strigins, Jevgenijs.
Proceedings of the 20th Nordic Conference of Computational Linguistics, NODALIDA 2015. Association for Computational Linguistics (ACL), 2015. p. 281-285 (Proceedings of the 20th Nordic Conference of Computational Linguistics, NODALIDA 2015).

Research output: Chapter in Book/Report/Conference proceeding › Conference paper › Research › peer-review

TY - GEN

T1 - Using sub-word n-gram models for dealing with OOV in large vocabulary speech recognition for Latvian

AU - Salimbajevs, Askars

AU - Strigins, Jevgenijs

PY - 2015

Y1 - 2015

N2 - In the Latvian language, one word can have tens or even hundreds of surface forms. This is a serious problem for large vocabulary speech recognition. Inclusion of every form in vocabulary will make it intractable, but, on the other hand, even with a vocabulary of 400K, the out-ofvocabulary (OOV) rate will be very high. In this paper, the authors investigate the possibility of using sub-word vocabularies where words are split into frequent and common parts. The results of our experiment show that this allows to significantly reduce the OOV rate.

AB - In the Latvian language, one word can have tens or even hundreds of surface forms. This is a serious problem for large vocabulary speech recognition. Inclusion of every form in vocabulary will make it intractable, but, on the other hand, even with a vocabulary of 400K, the out-ofvocabulary (OOV) rate will be very high. In this paper, the authors investigate the possibility of using sub-word vocabularies where words are split into frequent and common parts. The results of our experiment show that this allows to significantly reduce the OOV rate.

UR - https://www.scopus.com/pages/publications/84992577137

M3 - Conference paper

AN - SCOPUS:84992577137

SN - 9789175190983

T3 - Proceedings of the 20th Nordic Conference of Computational Linguistics, NODALIDA 2015

SP - 281

EP - 285

BT - Proceedings of the 20th Nordic Conference of Computational Linguistics, NODALIDA 2015

PB - Association for Computational Linguistics (ACL)

T2 - 20th Nordic Conference of Computational Linguistics, NODALIDA 2015

Y2 - 11 May 2015 through 13 May 2015

ER -

Salimbajevs A, Strigins J. Using sub-word n-gram models for dealing with OOV in large vocabulary speech recognition for Latvian. In Proceedings of the 20th Nordic Conference of Computational Linguistics, NODALIDA 2015. Association for Computational Linguistics (ACL). 2015. p. 281-285. (Proceedings of the 20th Nordic Conference of Computational Linguistics, NODALIDA 2015).

Using sub-word n-gram models for dealing with OOV in large vocabulary speech recognition for Latvian

Abstract

Publication series

Conference

OECD Field of Science

Other files and links

Fingerprint

Cite this