| 電子情報通信学会 研究会発表申込システム 講演論文 詳細 |
技報閲覧サービス [ログイン] 技報アーカイブ |
| トップに戻る | 前のページに戻る | [Japanese] / [English] |
| 講演抄録／キーワード | ||
| 講演名 | 2008-03-20 15:15 ［ポスター講演］Efficient Discriminative Training of Error Corrective Models Using High-WER Competitors ○Takanobu Oba・Takaaki Hori・Atsushi Nakamura（NTT）  SP2007-203 |
|
| 抄録 | （和） | We focus on error corrective models for speech recognition, which select a more accurate word sequence in a word N-best list produced by a speech recognizer. In general, an error corrective model is trained so that the correct word sequence is discriminated by the model from all the sequences in each N-best list. However, we show that a more accurate model can be provided through the discrimination of only two word sequences; one is the correct word sequence and the other is the sequence with the highest word error rate (WER) in the list. We think there are two reasons for this effect: (1) many important error patterns can be obtained since typical errors frequently appear in the word sequences with high WERs, and (2) training with fewer word sequences alleviates the difficulty of parameter estimation. In addition, the model size becomes more compact and the training time is significantly reduced with our method. Experiments using the Corpus of Spontaneous Japanese (CSJ) show that our proposed method generates accurate and compact models, and specifically performs robustly for different tasks and linguistic features. |
| （英） | We focus on error corrective models for speech recognition, which select a more accurate word sequence in a word N-best list produced by a speech recognizer. In general, an error corrective model is trained so that the correct word sequence is discriminated by the model from all the sequences in each N-best list. However, we show that a more accurate model can be provided through the discrimination of only two word sequences; one is the correct word sequence and the other is the sequence with the highest word error rate (WER) in the list. We think there are two reasons for this effect: (1) many important error patterns can be obtained since typical errors frequently appear in the word sequences with high WERs, and (2) training with fewer word sequences alleviates the difficulty of parameter estimation. In addition, the model size becomes more compact and the training time is significantly reduced with our method. Experiments using the Corpus of Spontaneous Japanese (CSJ) show that our proposed method generates accurate and compact models, and specifically performs robustly for different tasks and linguistic features. | |
| キーワード | （和） | error corrective model / high-WER competitors / / / / / / |
| （英） | error corrective model / high-WER competitors / / / / / / | |
| 文献情報 | 信学技報, vol. 107, no. 551, SP2007-203, pp. 99-104, 2008年3月. | |
| 資料番号 | SP2007-203 | |
| 発行日 | 2008-03-13 (SP) | |
| ISSN | Print edition: ISSN 0913-5685 Online edition: ISSN 2432-6380 | |
| 著作権に ついて |
技術研究報告に掲載された論文の著作権は電子情報通信学会に帰属します．(許諾番号：10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) | |
| PDFダウンロード | SP2007-203 |
|
| 研究会情報 | |
| 研究会 | SP |
| 開催期間 | 2008-03-20 - 2008-03-21 |
| 開催地（和） | 東大 |
| 開催地（英） | Univ. Tokyo |
| テーマ（和） | 国際ワークショップ(20日），研究会(21日) 「聴覚・音声・言語とその障害，一般」 |
| テーマ（英） | International Workshop (Mar 20), Speech Production, Speech Perception, Hearing and Speech, etc. (Mar 21) |
| 講演論文情報の詳細 | |
| 申込み研究会 | SP |
| 会議コード | 2008-03-SP |
| 本文の言語 | 英語 |
| タイトル（和） | |
| サブタイトル（和） | |
| タイトル（英） | Efficient Discriminative Training of Error Corrective Models Using High-WER Competitors |
| サブタイトル（英） | |
| キーワード(1)（和/英） | error corrective model / error corrective model |
| キーワード(2)（和/英） | high-WER competitors / high-WER competitors |
| キーワード(3)（和/英） | / |
| キーワード(4)（和/英） | / |
| キーワード(5)（和/英） | / |
| キーワード(6)（和/英） | / |
| キーワード(7)（和/英） | / |
| キーワード(8)（和/英） | / |
| 第1著者 氏名（和/英/ヨミ） | Takanobu Oba / Takanobu Oba / |
| 第1著者 所属（和/英） | NTT Communication Science Labs, NTT Corporation (略称： NTT) NTT Communication Science Labs, NTT Corporation (略称： NTT) |
| 第2著者 氏名（和/英/ヨミ） | Takaaki Hori / Takaaki Hori / |
| 第2著者 所属（和/英） | NTT Communication Science Labs, NTT Corporation (略称： NTT) NTT Communication Science Labs, NTT Corporation (略称： NTT) |
| 第3著者 氏名（和/英/ヨミ） | Atsushi Nakamura / Atsushi Nakamura / |
| 第3著者 所属（和/英） | NTT Communication Science Labs, NTT Corporation (略称： NTT) NTT Communication Science Labs, NTT Corporation (略称： NTT) |
| 講演者 | 第1著者 |
| 発表日時 | 2008-03-20 15:15:00 |
| 発表時間 | 90分 |
| 申込先研究会 | SP |
| 資料番号 | SP2007-203 |
| 巻番号（vol） | vol.107 |
| 号番号（no） | no.551 |
| ページ範囲 | pp.99-104 |
| ページ数 | 6 |
| 発行日 | 2008-03-13 (SP) |
[研究会発表申込システムのトップページに戻る]