Presentation 2009/12/14
Context-sensitive Statistical Models for Speaking-style Transformation
Graham Neubig, Yuya Akita, Shinsuke Mori, Tatsuya Kawahara,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Automatic speech recognition(ASR)results contain not only recognition errors, but also disfluencies and colloquial expressions that are not appropriate for inclusion in official transcripts. In order to correct these phenomena and create natural transcripts, we treat ASR results(or faithful transcripts)and official transcripts as different languages and use techniques from statistical machine translation(SMT)to "translate" between the two. In this paper, we present two novel methods in this framework. First, we introduce a technique to create context-sensitive translation models, improving the modeling accuracy. Second, we use log-linear interpolation to combine the translation model's joint and conditional probabilities, allowing for frequently observed patterns to be given higher priority. A system containing these improvements was implemented using weighted finite state transducers, and an evaluation was performed on transcripts from meetings of the Japanese Diet(national congress).
Keyword(in Japanese) (See Japanese page)
Keyword(in English)
Paper #
Date of Issue

Conference Information
Committee NLC
Conference Date 2009/12/14(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Natural Language Understanding and Models of Communication (NLC)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Context-sensitive Statistical Models for Speaking-style Transformation
Sub Title (in English)
Keyword(1)
1st Author's Name Graham Neubig
1st Author's Affiliation Graduate School of Informatics, Kyoto University()
2nd Author's Name Yuya Akita
2nd Author's Affiliation Graduate School of Informatics, Kyoto University
3rd Author's Name Shinsuke Mori
3rd Author's Affiliation Graduate School of Informatics, Kyoto University
4th Author's Name Tatsuya Kawahara
4th Author's Affiliation Graduate School of Informatics, Kyoto University
Date 2009/12/14
Paper #
Volume (vol) vol.109
Number (no) 355
Page pp.pp.-
#Pages 6
Date of Issue