Presentation 2013-06-14
Language Model Updating for Captioning of Disaster Broadcasts
Takahiro OKU, Yuya FUJITA, Akio KOBAYASHI, Shoei SATO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In this paper, we propose a language model (LM) adaptation method for transcribing disaster broadcasts. Because the disaster itself is an unprecedented event, there might be not enough quantities of training texts for the LM adaptation in the disaster. Then, our proposed method adapts LM in an unsupervised manner by using speech recognition results in addition to the online manuscripts. The proposed method, furthermore, employs a progressive updating procedure, which adapts in a short period to reflect the latest situation of the disaster to the LM. Experimental results showed that the proposed LM adaptation method reduced relatively a word error rate by 8.2% compared with the conventional LM adaptation method with online manuscripts only.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Unsupervised adaptation / Closed captioning / Disaster broadcasting
Paper # SP2013-48,WIT2013-18
Date of Issue

Conference Information
Committee WIT
Conference Date 2013/6/6(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 Well-being Information Technology(WIT)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Language Model Updating for Captioning of Disaster Broadcasts
Sub Title (in English)
Keyword(1) Unsupervised adaptation
Keyword(2) Closed captioning
Keyword(3) Disaster broadcasting
1st Author's Name Takahiro OKU
1st Author's Affiliation NHK Science and Technology Research Laboratories()
2nd Author's Name Yuya FUJITA
2nd Author's Affiliation NHK Science and Technology Research Laboratories
3rd Author's Name Akio KOBAYASHI
3rd Author's Affiliation NHK Science and Technology Research Laboratories
4th Author's Name Shoei SATO
4th Author's Affiliation NHK Science and Technology Research Laboratories
Date 2013-06-14
Paper # SP2013-48,WIT2013-18
Volume (vol) vol.113
Number (no) 77
Page pp.pp.-
#Pages 6
Date of Issue