A spoken term detection framework for recovering out-of-vocabulary words using the web

Carolina Parada; Abhinav Sethy; Mark Dredze; Frederick Jelinek

A spoken term detection framework for recovering out-of-vocabulary words using the web

Carolina Parada, Abhinav Sethy, Mark Dredze, Frederick Jelinek

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

Abstract

Vocabulary restrictions in large vocabulary continuous speech recognition (LVCSR) systems mean that out-of-vocabulary (OOV) words are lost in the output. However, OOV words tend to be information rich terms (often named entities) and their omission from the transcript negatively affects both usability and downstream NLP technologies, such as machine translation or knowledge distillation. We propose a novel approach to OOV recovery that uses a spoken term detection (STD) framework. Given an identified OOV region in the LVCSR output, we recover the uttered OOVs by utilizing contextual information and the vast and constantly updated vocabulary on the Web. Discovered words are integrated into system output, recovering up to 40% of OOVs and resulting in a reduction in system error.

Original language	English (US)
Title of host publication	Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010
Publisher	International Speech Communication Association
Pages	1269-1272
Number of pages	4
State	Published - 2010
Externally published	Yes

Publication series

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

Keywords

Data selection
Language modeling
Oov detection
Spoken term detection

ASJC Scopus subject areas

Language and Linguistics
Speech and Hearing
Human-Computer Interaction
Signal Processing
Software
Modeling and Simulation

Cite this

Parada, C., Sethy, A., Dredze, M., & Jelinek, F. (2010). A spoken term detection framework for recovering out-of-vocabulary words using the web. In Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010 (pp. 1269-1272). (Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010). International Speech Communication Association.

A spoken term detection framework for recovering out-of-vocabulary words using the web. / Parada, Carolina; Sethy, Abhinav; Dredze, Mark et al.
Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010. International Speech Communication Association, 2010. p. 1269-1272 (Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010).

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

Parada, C, Sethy, A, Dredze, M & Jelinek, F 2010, A spoken term detection framework for recovering out-of-vocabulary words using the web. in Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010. Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010, International Speech Communication Association, pp. 1269-1272.

Parada C, Sethy A, Dredze M, Jelinek F. A spoken term detection framework for recovering out-of-vocabulary words using the web. In Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010. International Speech Communication Association. 2010. p. 1269-1272. (Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010).

Parada, Carolina ; Sethy, Abhinav ; Dredze, Mark et al. / A spoken term detection framework for recovering out-of-vocabulary words using the web. Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010. International Speech Communication Association, 2010. pp. 1269-1272 (Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010).

@inproceedings{cf5b92a9f2834041ad53397fc9347aaa,

title = "A spoken term detection framework for recovering out-of-vocabulary words using the web",

abstract = "Vocabulary restrictions in large vocabulary continuous speech recognition (LVCSR) systems mean that out-of-vocabulary (OOV) words are lost in the output. However, OOV words tend to be information rich terms (often named entities) and their omission from the transcript negatively affects both usability and downstream NLP technologies, such as machine translation or knowledge distillation. We propose a novel approach to OOV recovery that uses a spoken term detection (STD) framework. Given an identified OOV region in the LVCSR output, we recover the uttered OOVs by utilizing contextual information and the vast and constantly updated vocabulary on the Web. Discovered words are integrated into system output, recovering up to 40% of OOVs and resulting in a reduction in system error.",

keywords = "Data selection, Language modeling, Oov detection, Spoken term detection",

author = "Carolina Parada and Abhinav Sethy and Mark Dredze and Frederick Jelinek",

note = "Funding Information: We thank Bhuvana Ramabhadran for many insightful discussions and the IBM Speech Group for the recognition lattices.",

year = "2010",

language = "English (US)",

series = "Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010",

publisher = "International Speech Communication Association",

pages = "1269--1272",

booktitle = "Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010",

}

TY - GEN

T1 - A spoken term detection framework for recovering out-of-vocabulary words using the web

AU - Parada, Carolina

AU - Sethy, Abhinav

AU - Dredze, Mark

AU - Jelinek, Frederick

N1 - Funding Information: We thank Bhuvana Ramabhadran for many insightful discussions and the IBM Speech Group for the recognition lattices.

PY - 2010

Y1 - 2010

N2 - Vocabulary restrictions in large vocabulary continuous speech recognition (LVCSR) systems mean that out-of-vocabulary (OOV) words are lost in the output. However, OOV words tend to be information rich terms (often named entities) and their omission from the transcript negatively affects both usability and downstream NLP technologies, such as machine translation or knowledge distillation. We propose a novel approach to OOV recovery that uses a spoken term detection (STD) framework. Given an identified OOV region in the LVCSR output, we recover the uttered OOVs by utilizing contextual information and the vast and constantly updated vocabulary on the Web. Discovered words are integrated into system output, recovering up to 40% of OOVs and resulting in a reduction in system error.

AB - Vocabulary restrictions in large vocabulary continuous speech recognition (LVCSR) systems mean that out-of-vocabulary (OOV) words are lost in the output. However, OOV words tend to be information rich terms (often named entities) and their omission from the transcript negatively affects both usability and downstream NLP technologies, such as machine translation or knowledge distillation. We propose a novel approach to OOV recovery that uses a spoken term detection (STD) framework. Given an identified OOV region in the LVCSR output, we recover the uttered OOVs by utilizing contextual information and the vast and constantly updated vocabulary on the Web. Discovered words are integrated into system output, recovering up to 40% of OOVs and resulting in a reduction in system error.

KW - Data selection

KW - Language modeling

KW - Oov detection

KW - Spoken term detection

UR - http://www.scopus.com/inward/record.url?scp=79959838750&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79959838750&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:79959838750

T3 - Proceedings of the 11th Annual Conference of the International Speech Communication Association, INTERSPEECH 2010

SP - 1269

EP - 1272

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

PB - International Speech Communication Association

ER -

A spoken term detection framework for recovering out-of-vocabulary words using the web

Abstract

Publication series

Keywords

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this