Comparison of Rule-based and Statistical Methods for Grapheme to Phoneme Modelling

Ilze Auziņa; Marcis Pinnis; Roberts Dargis

doi:10.3233/978-1-61499-442-8-57

Comparison of Rule-based and Statistical Methods for Grapheme to Phoneme Modelling

Ilze Auziņa^*
, Marcis Pinnis
, Roberts Dargis

^*Corresponding author for this work

University of Latvia
Tilde Company

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

3 Citations (Scopus)

Abstract

Grapheme to phoneme modelling is one of the key features in automated speech recognition and speech synthesis. In this paper, the authors compare two different approaches: a statistical machine translation based method using the phonetically transcribed Latvian Speech Recognition Corpus and a rule-based method for phonetic transcription of words from grammatically correct forms. The paper provides 10-fold cross-validation results and error analysis for both methods.

Original language	English
Title of host publication	Human Language Technologies - The Baltic Perspective
Subtitle of host publication	Proceedings of the 6th International Conference Baltic HLT 2014
Editors	Andrius Utka, Gintare Grigonyte, Jurgita Kapociute-Dzikiene, Jurgita Vaicenoniene
Publisher	IOS Press BV
Pages	57-60
Number of pages	4
ISBN (Electronic)	9781614994411
DOIs	https://doi.org/10.3233/978-1-61499-442-8-57
Publication status	Published - 2014
Externally published	Yes
Event	6th International Conference on Human Language Technologies - The Baltic Perspective, Baltic HLT 2014 - Kaunas, Lithuania Duration: 26 Sept 2014 → 27 Sept 2014

Publication series

Name	Frontiers in Artificial Intelligence and Applications
Volume	268
ISSN (Print)	0922-6389
ISSN (Electronic)	1879-8314

Conference

Conference	6th International Conference on Human Language Technologies - The Baltic Perspective, Baltic HLT 2014
Country/Territory	Lithuania
City	Kaunas
Period	26/09/14 → 27/09/14

Keywords

Grapheme to phoneme modelling
Latvian language
method comparison

Access to Document

10.3233/978-1-61499-442-8-57

Cite this

Auziņa, I., Pinnis, M., & Dargis, R. (2014). Comparison of Rule-based and Statistical Methods for Grapheme to Phoneme Modelling. In A. Utka, G. Grigonyte, J. Kapociute-Dzikiene, & J. Vaicenoniene (Eds.), Human Language Technologies - The Baltic Perspective: Proceedings of the 6th International Conference Baltic HLT 2014 (pp. 57-60). (Frontiers in Artificial Intelligence and Applications; Vol. 268). IOS Press BV. https://doi.org/10.3233/978-1-61499-442-8-57

Auziņa, Ilze ; Pinnis, Marcis ; Dargis, Roberts. / Comparison of Rule-based and Statistical Methods for Grapheme to Phoneme Modelling. Human Language Technologies - The Baltic Perspective: Proceedings of the 6th International Conference Baltic HLT 2014. editor / Andrius Utka ; Gintare Grigonyte ; Jurgita Kapociute-Dzikiene ; Jurgita Vaicenoniene. IOS Press BV, 2014. pp. 57-60 (Frontiers in Artificial Intelligence and Applications).

@inproceedings{a1d68eb40cbf40bf879848639d082f27,

title = "Comparison of Rule-based and Statistical Methods for Grapheme to Phoneme Modelling",

abstract = "Grapheme to phoneme modelling is one of the key features in automated speech recognition and speech synthesis. In this paper, the authors compare two different approaches: a statistical machine translation based method using the phonetically transcribed Latvian Speech Recognition Corpus and a rule-based method for phonetic transcription of words from grammatically correct forms. The paper provides 10-fold cross-validation results and error analysis for both methods.",

keywords = "Grapheme to phoneme modelling, Latvian language, method comparison",

author = "Ilze Auziņa and Marcis Pinnis and Roberts Dargis",

note = "Publisher Copyright: {\textcopyright} 2014 The Authors and IOS Press.; 6th International Conference on Human Language Technologies - The Baltic Perspective, Baltic HLT 2014 ; Conference date: 26-09-2014 Through 27-09-2014",

year = "2014",

doi = "10.3233/978-1-61499-442-8-57",

language = "English",

series = "Frontiers in Artificial Intelligence and Applications",

publisher = "IOS Press BV",

pages = "57--60",

editor = "Andrius Utka and Gintare Grigonyte and Jurgita Kapociute-Dzikiene and Jurgita Vaicenoniene",

booktitle = "Human Language Technologies - The Baltic Perspective",

address = "Netherlands",

}

Auziņa, I, Pinnis, M & Dargis, R 2014, Comparison of Rule-based and Statistical Methods for Grapheme to Phoneme Modelling. in A Utka, G Grigonyte, J Kapociute-Dzikiene & J Vaicenoniene (eds), Human Language Technologies - The Baltic Perspective: Proceedings of the 6th International Conference Baltic HLT 2014. Frontiers in Artificial Intelligence and Applications, vol. 268, IOS Press BV, pp. 57-60, 6th International Conference on Human Language Technologies - The Baltic Perspective, Baltic HLT 2014, Kaunas, Lithuania, 26/09/14. https://doi.org/10.3233/978-1-61499-442-8-57

Comparison of Rule-based and Statistical Methods for Grapheme to Phoneme Modelling. / Auziņa, Ilze; Pinnis, Marcis; Dargis, Roberts.
Human Language Technologies - The Baltic Perspective: Proceedings of the 6th International Conference Baltic HLT 2014. ed. / Andrius Utka; Gintare Grigonyte; Jurgita Kapociute-Dzikiene; Jurgita Vaicenoniene. IOS Press BV, 2014. p. 57-60 (Frontiers in Artificial Intelligence and Applications; Vol. 268).

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

TY - GEN

T1 - Comparison of Rule-based and Statistical Methods for Grapheme to Phoneme Modelling

AU - Auziņa, Ilze

AU - Pinnis, Marcis

AU - Dargis, Roberts

PY - 2014

Y1 - 2014

N2 - Grapheme to phoneme modelling is one of the key features in automated speech recognition and speech synthesis. In this paper, the authors compare two different approaches: a statistical machine translation based method using the phonetically transcribed Latvian Speech Recognition Corpus and a rule-based method for phonetic transcription of words from grammatically correct forms. The paper provides 10-fold cross-validation results and error analysis for both methods.

AB - Grapheme to phoneme modelling is one of the key features in automated speech recognition and speech synthesis. In this paper, the authors compare two different approaches: a statistical machine translation based method using the phonetically transcribed Latvian Speech Recognition Corpus and a rule-based method for phonetic transcription of words from grammatically correct forms. The paper provides 10-fold cross-validation results and error analysis for both methods.

KW - Grapheme to phoneme modelling

KW - Latvian language

KW - method comparison

UR - https://www.scopus.com/pages/publications/84948660613

U2 - 10.3233/978-1-61499-442-8-57

DO - 10.3233/978-1-61499-442-8-57

M3 - Conference paper

AN - SCOPUS:84948660613

T3 - Frontiers in Artificial Intelligence and Applications

SP - 57

EP - 60

BT - Human Language Technologies - The Baltic Perspective

A2 - Utka, Andrius

A2 - Grigonyte, Gintare

A2 - Kapociute-Dzikiene, Jurgita

A2 - Vaicenoniene, Jurgita

PB - IOS Press BV

T2 - 6th International Conference on Human Language Technologies - The Baltic Perspective, Baltic HLT 2014

Y2 - 26 September 2014 through 27 September 2014

ER -

Auziņa I, Pinnis M, Dargis R. Comparison of Rule-based and Statistical Methods for Grapheme to Phoneme Modelling. In Utka A, Grigonyte G, Kapociute-Dzikiene J, Vaicenoniene J, editors, Human Language Technologies - The Baltic Perspective: Proceedings of the 6th International Conference Baltic HLT 2014. IOS Press BV. 2014. p. 57-60. (Frontiers in Artificial Intelligence and Applications). doi: 10.3233/978-1-61499-442-8-57

Comparison of Rule-based and Statistical Methods for Grapheme to Phoneme Modelling

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this