Modeling pronunciation variation for a dutch CSR: testing three methods

Mirjam Wester; Judith M. Kessens; Helmer Strik

Modeling pronunciation variation for a dutch CSR: testing three methods

Mirjam Wester, Judith M. Kessens, Helmer Strik

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract / Description of output

This paper describes how the performance of a continuous speech recognizer for Dutch has been improved by modeling pronunciation variation. We used three methods to model pronunciation variation. First, within-word variation was dealt with. Phonological rules were applied to the words in the lexicon, thus automatically generating pronunciation variants. Secondly, cross-word pronunciation variation was modeled using two different approaches. The first approach was to model cross-word processes by adding the variants as separate words to the lexicon and in the second approach this was done by using multi-words. For each of the methods, recognition experiments were carried out. A significant improvement was found for modeling within-word variation. Furthermore, modeling cross-word processes using multi-words leads to significantly better results than modeling them using separate words in the lexicon.

Original language	English
Title of host publication	The 5th International Conference on Spoken Language Processing, Incorporating The 7th Australian International Speech Science and Technology Conference, Sydney Convention Centre, Sydney, Australia, 30th November - 4th December 1998
Number of pages	4
Publication status	Published - 1998

Access to Document

i98_0371Final published version, 115 KB

http://www.isca-speech.org/archive/icslp_1998/i98_0371.html

Cite this

Wester, M., Kessens, J. M., & Strik, H. (1998). Modeling pronunciation variation for a dutch CSR: testing three methods. In The 5th International Conference on Spoken Language Processing, Incorporating The 7th Australian International Speech Science and Technology Conference, Sydney Convention Centre, Sydney, Australia, 30th November - 4th December 1998 Article 0371 http://www.isca-speech.org/archive/icslp_1998/i98_0371.html

@inproceedings{b2f9c741eb154a1bbccc764fa75ec62d,

title = "Modeling pronunciation variation for a dutch CSR: testing three methods",

abstract = "This paper describes how the performance of a continuous speech recognizer for Dutch has been improved by modeling pronunciation variation. We used three methods to model pronunciation variation. First, within-word variation was dealt with. Phonological rules were applied to the words in the lexicon, thus automatically generating pronunciation variants. Secondly, cross-word pronunciation variation was modeled using two different approaches. The first approach was to model cross-word processes by adding the variants as separate words to the lexicon and in the second approach this was done by using multi-words. For each of the methods, recognition experiments were carried out. A significant improvement was found for modeling within-word variation. Furthermore, modeling cross-word processes using multi-words leads to significantly better results than modeling them using separate words in the lexicon.",

author = "Mirjam Wester and Kessens, {Judith M.} and Helmer Strik",

year = "1998",

language = "English",

booktitle = "The 5th International Conference on Spoken Language Processing, Incorporating The 7th Australian International Speech Science and Technology Conference, Sydney Convention Centre, Sydney, Australia, 30th November - 4th December 1998",

}

Wester, M, Kessens, JM & Strik, H 1998, Modeling pronunciation variation for a dutch CSR: testing three methods. in The 5th International Conference on Spoken Language Processing, Incorporating The 7th Australian International Speech Science and Technology Conference, Sydney Convention Centre, Sydney, Australia, 30th November - 4th December 1998., 0371. <http://www.isca-speech.org/archive/icslp_1998/i98_0371.html>

Modeling pronunciation variation for a dutch CSR: testing three methods. / Wester, Mirjam; Kessens, Judith M.; Strik, Helmer.
The 5th International Conference on Spoken Language Processing, Incorporating The 7th Australian International Speech Science and Technology Conference, Sydney Convention Centre, Sydney, Australia, 30th November - 4th December 1998. 1998. 0371.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - Modeling pronunciation variation for a dutch CSR: testing three methods

AU - Wester, Mirjam

AU - Kessens, Judith M.

AU - Strik, Helmer

PY - 1998

Y1 - 1998

N2 - This paper describes how the performance of a continuous speech recognizer for Dutch has been improved by modeling pronunciation variation. We used three methods to model pronunciation variation. First, within-word variation was dealt with. Phonological rules were applied to the words in the lexicon, thus automatically generating pronunciation variants. Secondly, cross-word pronunciation variation was modeled using two different approaches. The first approach was to model cross-word processes by adding the variants as separate words to the lexicon and in the second approach this was done by using multi-words. For each of the methods, recognition experiments were carried out. A significant improvement was found for modeling within-word variation. Furthermore, modeling cross-word processes using multi-words leads to significantly better results than modeling them using separate words in the lexicon.

AB - This paper describes how the performance of a continuous speech recognizer for Dutch has been improved by modeling pronunciation variation. We used three methods to model pronunciation variation. First, within-word variation was dealt with. Phonological rules were applied to the words in the lexicon, thus automatically generating pronunciation variants. Secondly, cross-word pronunciation variation was modeled using two different approaches. The first approach was to model cross-word processes by adding the variants as separate words to the lexicon and in the second approach this was done by using multi-words. For each of the methods, recognition experiments were carried out. A significant improvement was found for modeling within-word variation. Furthermore, modeling cross-word processes using multi-words leads to significantly better results than modeling them using separate words in the lexicon.

M3 - Conference contribution

BT - The 5th International Conference on Spoken Language Processing, Incorporating The 7th Australian International Speech Science and Technology Conference, Sydney Convention Centre, Sydney, Australia, 30th November - 4th December 1998

ER -

Modeling pronunciation variation for a dutch CSR: testing three methods

Abstract / Description of output

Access to Document

Fingerprint

Cite this