Analysis of Phonetic Transcriptions for Danish Automatic Speech Recognition

Andreas Søeborg Kirkedal

Analysis of Phonetic Transcriptions for Danish Automatic Speech Recognition

Andreas Søeborg Kirkedal

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

Abstract

Automatic speech recognition (ASR) relies on three resources: audio, orthographic transcriptions and a pronunciation dictionary. The dictionary or lexicon maps orthographic words to sequences of phones or phonemes that represent the pronunciation of the corresponding word. The quality of a speech recognition system depends heavily on the dictionary and the transcriptions therein. This paper presents an analysis of phonetic/phonemic features that are salient for current Danish ASR systems. This preliminary study consists of a series of experiments using an ASR system trained on the DK-PAROLE corpus. The analysis indicates that transcribing e.g. stress or vowel duration has a negative impact on performance. The best performance is obtained with coarse phonetic annotation and improves performance 1% word error rate and 3.8% sentence error rate.

Original language	English
Title of host publication	Proceedings of the 19th Nordic Conference of Computational Linguistics (NODALIDA 2013)
Editors	Stephan Oepen, Kristin Hagen, Janne Bondi Johannessen
Place of Publication	Linköping
Publisher	Linköping University Electronic Press
Publication date	2013
Pages	321-330
ISBN (Print)	9789175195896
Publication status	Published - 2013
Event	NODALIDA 2013: The 19th Nordic Conference of Computational Linguistics - University of Oslo, Oslo, Norway Duration: 22 May 2013 → 24 May 2013 Conference number: 19 http://www.hf.uio.no/iln/english/research/news-and-events/events/conferences/2013/nodalida/index.html

Conference

Conference	NODALIDA 2013
Number	19
Location	University of Oslo
Country/Territory	Norway
City	Oslo
Period	22/05/2013 → 24/05/2013
Internet address	http://www.hf.uio.no/iln/english/research/news-and-events/events/conferences/2013/nodalida/index.html

Series	NEALT (Northern European Association of Language Technology) Proceedings Series
Volume	16
ISSN	1736-6305

Access to Document

http://www.ep.liu.se/ecp_home/index.en.aspx?issue=085

CBS Remote Access

Full text

Handle.net

Persistent link

Cite this

Kirkedal, A. S. (2013). Analysis of Phonetic Transcriptions for Danish Automatic Speech Recognition. In S. Oepen, K. Hagen, & J. B. Johannessen (Eds.), Proceedings of the 19th Nordic Conference of Computational Linguistics (NODALIDA 2013) (pp. 321-330). Linköping University Electronic Press. NEALT (Northern European Association of Language Technology) Proceedings Series Vol. 16 http://www.ep.liu.se/ecp_home/index.en.aspx?issue=085

Kirkedal, Andreas Søeborg. / Analysis of Phonetic Transcriptions for Danish Automatic Speech Recognition. Proceedings of the 19th Nordic Conference of Computational Linguistics (NODALIDA 2013). editor / Stephan Oepen ; Kristin Hagen ; Janne Bondi Johannessen. Linköping : Linköping University Electronic Press, 2013. pp. 321-330 (NEALT (Northern European Association of Language Technology) Proceedings Series, Vol. 16).

@inproceedings{29d081897a824b018104d4fc871eb849,

title = "Analysis of Phonetic Transcriptions for Danish Automatic Speech Recognition",

abstract = "Automatic speech recognition (ASR) relies on three resources: audio, orthographic transcriptions and a pronunciation dictionary. The dictionary or lexicon maps orthographic words to sequences of phones or phonemes that represent the pronunciation of the corresponding word. The quality of a speech recognition system depends heavily on the dictionary and the transcriptions therein. This paper presents an analysis of phonetic/phonemic features that are salient for current Danish ASR systems. This preliminary study consists of a series of experiments using an ASR system trained on the DK-PAROLE corpus. The analysis indicates that transcribing e.g. stress or vowel duration has a negative impact on performance. The best performance is obtained with coarse phonetic annotation and improves performance 1% word error rate and 3.8% sentence error rate.",

keywords = "Automatic speech recognition, Phonetics, Phonology, Speech, Phonetic transcription",

author = "Kirkedal, {Andreas S{\o}eborg}",

year = "2013",

language = "English",

isbn = "9789175195896",

series = "NEALT (Northern European Association of Language Technology) Proceedings Series",

pages = "321--330",

editor = "Oepen, {Stephan } and Hagen, {Kristin } and Johannessen, {Janne Bondi }",

booktitle = "Proceedings of the 19th Nordic Conference of Computational Linguistics (NODALIDA 2013)",

publisher = "Link{\"o}ping University Electronic Press",

address = "Sweden",

note = "null ; Conference date: 22-05-2013 Through 24-05-2013",

url = "http://www.hf.uio.no/iln/english/research/news-and-events/events/conferences/2013/nodalida/index.html",

}

Kirkedal, AS 2013, Analysis of Phonetic Transcriptions for Danish Automatic Speech Recognition. in S Oepen, K Hagen & JB Johannessen (eds), Proceedings of the 19th Nordic Conference of Computational Linguistics (NODALIDA 2013). Linköping University Electronic Press, Linköping, NEALT (Northern European Association of Language Technology) Proceedings Series, vol. 16, pp. 321-330, NODALIDA 2013, Oslo, Norway, 22/05/2013. <http://www.ep.liu.se/ecp_home/index.en.aspx?issue=085>

Analysis of Phonetic Transcriptions for Danish Automatic Speech Recognition. / Kirkedal, Andreas Søeborg.

Proceedings of the 19th Nordic Conference of Computational Linguistics (NODALIDA 2013). ed. / Stephan Oepen; Kristin Hagen; Janne Bondi Johannessen. Linköping : Linköping University Electronic Press, 2013. p. 321-330 (NEALT (Northern European Association of Language Technology) Proceedings Series, Vol. 16).

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

TY - GEN

T1 - Analysis of Phonetic Transcriptions for Danish Automatic Speech Recognition

AU - Kirkedal, Andreas Søeborg

N1 - Conference code: 19

PY - 2013

Y1 - 2013

N2 - Automatic speech recognition (ASR) relies on three resources: audio, orthographic transcriptions and a pronunciation dictionary. The dictionary or lexicon maps orthographic words to sequences of phones or phonemes that represent the pronunciation of the corresponding word. The quality of a speech recognition system depends heavily on the dictionary and the transcriptions therein. This paper presents an analysis of phonetic/phonemic features that are salient for current Danish ASR systems. This preliminary study consists of a series of experiments using an ASR system trained on the DK-PAROLE corpus. The analysis indicates that transcribing e.g. stress or vowel duration has a negative impact on performance. The best performance is obtained with coarse phonetic annotation and improves performance 1% word error rate and 3.8% sentence error rate.

AB - Automatic speech recognition (ASR) relies on three resources: audio, orthographic transcriptions and a pronunciation dictionary. The dictionary or lexicon maps orthographic words to sequences of phones or phonemes that represent the pronunciation of the corresponding word. The quality of a speech recognition system depends heavily on the dictionary and the transcriptions therein. This paper presents an analysis of phonetic/phonemic features that are salient for current Danish ASR systems. This preliminary study consists of a series of experiments using an ASR system trained on the DK-PAROLE corpus. The analysis indicates that transcribing e.g. stress or vowel duration has a negative impact on performance. The best performance is obtained with coarse phonetic annotation and improves performance 1% word error rate and 3.8% sentence error rate.

KW - Automatic speech recognition

KW - Phonetics

KW - Phonology

KW - Speech

KW - Phonetic transcription

M3 - Article in proceedings

SN - 9789175195896

T3 - NEALT (Northern European Association of Language Technology) Proceedings Series

SP - 321

EP - 330

BT - Proceedings of the 19th Nordic Conference of Computational Linguistics (NODALIDA 2013)

A2 - Oepen, Stephan

A2 - Hagen, Kristin

A2 - Johannessen, Janne Bondi

PB - Linköping University Electronic Press

CY - Linköping

Y2 - 22 May 2013 through 24 May 2013

ER -

Kirkedal AS. Analysis of Phonetic Transcriptions for Danish Automatic Speech Recognition. In Oepen S, Hagen K, Johannessen JB, editors, Proceedings of the 19th Nordic Conference of Computational Linguistics (NODALIDA 2013). Linköping: Linköping University Electronic Press. 2013. p. 321-330. (NEALT (Northern European Association of Language Technology) Proceedings Series, Vol. 16).