ダウンロード数: 318
このアイテムのファイル:
ファイル | 記述 | サイズ | フォーマット | |
---|---|---|---|---|
TASLP.2016.2562505.pdf | 1.68 MB | Adobe PDF | 見る/開く |
タイトル: | Semi-Supervised Acoustic Model Training by Discriminative Data Selection from Multiple ASR Systems' Hypotheses |
著者: | Li, Sheng Akita, Yuya Kawahara, Tatsuya |
著者名の別形: | 河原, 達也 |
キーワード: | speech recognition acoustic model semi-supervised training lecture transcription |
発行日: | 3-May-2016 |
出版者: | Institute of Electrical and Electronics Engineers Inc. |
誌名: | IEEE/ACM Transactions on Audio Speech and Language Processing |
巻: | 24 |
号: | 9 |
開始ページ: | 1524 |
終了ページ: | 1534 |
抄録: | While the performance of ASR systems depends on the size of the training data, it is very costly to prepare accurate and faithful transcripts. In this paper, we investigate a semisupervised training scheme, which takes the advantage of huge quantities of unlabeled video lecture archive, particularly for the deep neural network (DNN) acoustic model. In the proposed method, we obtain ASR hypotheses by complementary GMM-and DNN-based ASR systems. Then, a set of CRF-based classifiers is trained to select the correct hypotheses and verify the selected data. The proposed hypothesis combination shows higher quality compared with the conventional system combination method (ROVER). Moreover, compared with the conventional data selection based on confidence measure score, our method is demonstrated more effective for filtering usable data. Significant improvement in the ASR accuracy is achieved over the baseline system and in comparison with the models trained with the conventional system combination and data selection methods. |
著作権等: | ©2016 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other users, including reprinting/ republishing this material for advertising or promotional purposes, creating new collective works for resale or redistribution to servers or lists, or reuse of any copyrighted components of this work in other works. この論文は出版社版でありません。引用の際には出版社版をご確認ご利用ください。 This is not the published version. Please cite only the published version. |
URI: | http://hdl.handle.net/2433/219417 |
DOI(出版社版): | 10.1109/TASLP.2016.2562505 |
出現コレクション: | 学術雑誌掲載論文等 |
このリポジトリに保管されているアイテムはすべて著作権により保護されています。