A speech understanding framework that uses multiple language models and multiple understanding models

Masaki Katsumaru; Mikio Nakano; Kazunori Komatani; Kotaro Funakoshi; Tetsuya Ogata; Hiroshi G. Okuno

A speech understanding framework that uses multiple language models and multiple understanding models

Masaki Katsumaru, Mikio Nakano, Kazunori Komatani, Kotaro Funakoshi, Tetsuya Ogata, Hiroshi G. Okuno

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

Abstract

The optimal combination of language model (LM) and language understanding model (LUM) varies depending on available training data and utterances to be handled. Usually, a lot of effort and time are needed to find the optimal combination. Instead, we have designed and developed a new framework that uses multiple LMs and LUMs to improve speech understanding accuracy under various situations. As one implementation of the framework, we have developed a method for selecting the most appropriate speech understanding result from several candidates. We use two LMs and three LUMs, and thus obtain six combinations of them. We empirically show that our method improves speech understanding accuracy. The performance of the oracle selection suggests further potential improvements in our system.

Original language	English
Title of host publication	NAACL-HLT 2009 - Human Language Technologies
Subtitle of host publication	2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Short Papers
Editors	Mari Ostendorf, Michael Collins, Shri Narayanan, Douglas W. Oard, Lucy Vanderwende
Publisher	Association for Computational Linguistics (ACL)
Pages	133-136
Number of pages	4
ISBN (Electronic)	9781932432428
Publication status	Published - 2009
Externally published	Yes
Event	2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL-HLT 2009 - Boulder, United States Duration: 2009 May 31 → 2009 Jun 5

Publication series

Name	NAACL-HLT 2009 - Human Language Technologies: 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Short Papers

Conference

Conference	2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL-HLT 2009
Country/Territory	United States
City	Boulder
Period	09/5/31 → 09/6/5

ASJC Scopus subject areas

Language and Linguistics
Linguistics and Language

Cite this

Katsumaru, M., Nakano, M., Komatani, K., Funakoshi, K., Ogata, T., & Okuno, H. G. (2009). A speech understanding framework that uses multiple language models and multiple understanding models. In M. Ostendorf, M. Collins, S. Narayanan, D. W. Oard, & L. Vanderwende (Eds.), NAACL-HLT 2009 - Human Language Technologies: 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Short Papers (pp. 133-136). (NAACL-HLT 2009 - Human Language Technologies: 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Short Papers). Association for Computational Linguistics (ACL).

A speech understanding framework that uses multiple language models and multiple understanding models. / Katsumaru, Masaki; Nakano, Mikio; Komatani, Kazunori et al.

NAACL-HLT 2009 - Human Language Technologies: 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Short Papers. ed. / Mari Ostendorf; Michael Collins; Shri Narayanan; Douglas W. Oard; Lucy Vanderwende. Association for Computational Linguistics (ACL), 2009. p. 133-136 (NAACL-HLT 2009 - Human Language Technologies: 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Short Papers).

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

Katsumaru, M, Nakano, M, Komatani, K, Funakoshi, K, Ogata, T & Okuno, HG 2009, A speech understanding framework that uses multiple language models and multiple understanding models. in M Ostendorf, M Collins, S Narayanan, DW Oard & L Vanderwende (eds), NAACL-HLT 2009 - Human Language Technologies: 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Short Papers. NAACL-HLT 2009 - Human Language Technologies: 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Short Papers, Association for Computational Linguistics (ACL), pp. 133-136, 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL-HLT 2009, Boulder, United States, 09/5/31.

Katsumaru M, Nakano M, Komatani K, Funakoshi K, Ogata T, Okuno HG. A speech understanding framework that uses multiple language models and multiple understanding models. In Ostendorf M, Collins M, Narayanan S, Oard DW, Vanderwende L, editors, NAACL-HLT 2009 - Human Language Technologies: 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Short Papers. Association for Computational Linguistics (ACL). 2009. p. 133-136. (NAACL-HLT 2009 - Human Language Technologies: 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Short Papers).

Katsumaru, Masaki ; Nakano, Mikio ; Komatani, Kazunori et al. / A speech understanding framework that uses multiple language models and multiple understanding models. NAACL-HLT 2009 - Human Language Technologies: 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Short Papers. editor / Mari Ostendorf ; Michael Collins ; Shri Narayanan ; Douglas W. Oard ; Lucy Vanderwende. Association for Computational Linguistics (ACL), 2009. pp. 133-136 (NAACL-HLT 2009 - Human Language Technologies: 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Short Papers).

@inproceedings{64c89f7fe4fb4d8f8337220820974fdb,

title = "A speech understanding framework that uses multiple language models and multiple understanding models",

abstract = "The optimal combination of language model (LM) and language understanding model (LUM) varies depending on available training data and utterances to be handled. Usually, a lot of effort and time are needed to find the optimal combination. Instead, we have designed and developed a new framework that uses multiple LMs and LUMs to improve speech understanding accuracy under various situations. As one implementation of the framework, we have developed a method for selecting the most appropriate speech understanding result from several candidates. We use two LMs and three LUMs, and thus obtain six combinations of them. We empirically show that our method improves speech understanding accuracy. The performance of the oracle selection suggests further potential improvements in our system.",

author = "Masaki Katsumaru and Mikio Nakano and Kazunori Komatani and Kotaro Funakoshi and Tetsuya Ogata and Okuno, {Hiroshi G.}",

note = "Publisher Copyright: {\textcopyright} 2009 Association for Computational Linguistics; 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL-HLT 2009 ; Conference date: 31-05-2009 Through 05-06-2009",

year = "2009",

language = "English",

series = "NAACL-HLT 2009 - Human Language Technologies: 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Short Papers",

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

pages = "133--136",

editor = "Mari Ostendorf and Michael Collins and Shri Narayanan and Oard, {Douglas W.} and Lucy Vanderwende",

booktitle = "NAACL-HLT 2009 - Human Language Technologies",

}

TY - GEN

T1 - A speech understanding framework that uses multiple language models and multiple understanding models

AU - Katsumaru, Masaki

AU - Nakano, Mikio

AU - Komatani, Kazunori

AU - Funakoshi, Kotaro

AU - Ogata, Tetsuya

AU - Okuno, Hiroshi G.

PY - 2009

Y1 - 2009

N2 - The optimal combination of language model (LM) and language understanding model (LUM) varies depending on available training data and utterances to be handled. Usually, a lot of effort and time are needed to find the optimal combination. Instead, we have designed and developed a new framework that uses multiple LMs and LUMs to improve speech understanding accuracy under various situations. As one implementation of the framework, we have developed a method for selecting the most appropriate speech understanding result from several candidates. We use two LMs and three LUMs, and thus obtain six combinations of them. We empirically show that our method improves speech understanding accuracy. The performance of the oracle selection suggests further potential improvements in our system.

AB - The optimal combination of language model (LM) and language understanding model (LUM) varies depending on available training data and utterances to be handled. Usually, a lot of effort and time are needed to find the optimal combination. Instead, we have designed and developed a new framework that uses multiple LMs and LUMs to improve speech understanding accuracy under various situations. As one implementation of the framework, we have developed a method for selecting the most appropriate speech understanding result from several candidates. We use two LMs and three LUMs, and thus obtain six combinations of them. We empirically show that our method improves speech understanding accuracy. The performance of the oracle selection suggests further potential improvements in our system.

UR - http://www.scopus.com/inward/record.url?scp=79959846818&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79959846818&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:79959846818

T3 - NAACL-HLT 2009 - Human Language Technologies: 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics, Short Papers

SP - 133

EP - 136

BT - NAACL-HLT 2009 - Human Language Technologies

A2 - Ostendorf, Mari

A2 - Collins, Michael

A2 - Narayanan, Shri

A2 - Oard, Douglas W.

A2 - Vanderwende, Lucy

PB - Association for Computational Linguistics (ACL)

T2 - 2009 Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, NAACL-HLT 2009

Y2 - 31 May 2009 through 5 June 2009

ER -

A speech understanding framework that uses multiple language models and multiple understanding models

Abstract

Publication series

Conference

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this