{"created":"2025-01-18T23:20:25.853366+00:00","updated":"2025-01-22T04:36:58.611335+00:00","metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00057169","sets":["1164:5159:5192:5193"]},"path":["5193"],"owner":"1","recid":"57169","title":["複数の雑音抑圧手法を用いた認識結果の統合による　ロバスト音声認識の検討"],"pubdate":{"attribute_name":"公開日","attribute_value":"2003-12-18"},"_buckets":{"deposit":"368dd634-d4b0-494d-8a48-30e4811a243c"},"_deposit":{"id":"57169","pid":{"type":"depid","value":"57169","revision_id":0},"owners":[1],"status":"published","created_by":1},"item_title":"複数の雑音抑圧手法を用いた認識結果の統合による　ロバスト音声認識の検討","author_link":["0","0"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"複数の雑音抑圧手法を用いた認識結果の統合による　ロバスト音声認識の検討"},{"subitem_title":"Integration of Recognition Results for Robust Speech Recognition","subitem_title_language":"en"}]},"item_type_id":"4","publish_date":"2003-12-18","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"筑波大学大学院システム情報工学研究科"},{"subitem_text_value":"筑波大学電子・情報工学系"},{"subitem_text_value":"筑波大学電子・情報工学系"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Graduate School of Systems and Information Engineering, University of Tsukuba","subitem_text_language":"en"},{"subitem_text_value":"Institute of Information Sciences and Electronics, University of Tsukuba","subitem_text_language":"en"},{"subitem_text_value":"Institute of Information Sciences and Electronics, University of Tsukuba","subitem_text_language":"en"}]},"item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"jpn"}]},"item_publisher":{"attribute_name":"出版者","attribute_value_mlt":[{"subitem_publisher":"情報処理学会","subitem_publisher_language":"ja"}]},"publish_status":"0","weko_shared_id":-1,"item_file_price":{"attribute_name":"Billing file","attribute_type":"file","attribute_value_mlt":[{"url":{"url":"https://ipsj.ixsq.nii.ac.jp/record/57169/files/IPSJ-SLP03049019.pdf"},"date":[{"dateType":"Available","dateValue":"2005-12-18"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-SLP03049019.pdf","filesize":[{"value":"473.2 kB"}],"mimetype":"application/pdf","priceinfo":[{"tax":["include_tax"],"price":"660","billingrole":"5"},{"tax":["include_tax"],"price":"330","billingrole":"6"},{"tax":["include_tax"],"price":"0","billingrole":"22"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_date","version_id":"5299b918-04ea-4e20-9141-1c3ed43df7c1","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2003 by the Information Processing Society of Japan"}]},"item_4_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"岡田, 治郎"},{"creatorName":"山田, 武志"},{"creatorName":"北脇, 信彦"}],"nameIdentifiers":[{}]}]},"item_4_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Jiro, Okada","creatorNameLang":"en"},{"creatorName":"Takeshi, Yamada","creatorNameLang":"en"},{"creatorName":"Nobuhiko, Kitawaki","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_4_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AN10442647","subitem_source_identifier_type":"NCID"}]},"item_4_textarea_12":{"attribute_name":"Notice","attribute_value_mlt":[{"subitem_textarea_value":"SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc."}]},"item_resource_type":{"attribute_name":"資源タイプ","attribute_value_mlt":[{"resourceuri":"http://purl.org/coar/resource_type/c_18gh","resourcetype":"technical report"}]},"item_4_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"様々な雑音条件に対してロバストな音声認識を実現するためには，複数の雑音抑圧手法の各々の認識結果から信頼度の高いものを選択する方法が有効であると考えられる．本稿では，フレーム正規化対数尤度に基づく信頼度を用いた認識結果の統合法を提案する．提案法では，正解と最もマッチする認識結果を得るために，各雑音抑圧手法を用いた時のN-bestの認識結果を求める．そして，各認識結果に対して信頼度を付与し，信頼度が最大になるものを選択する．提案法の有効性を評価するために，雑音下連続数字認識タスクであるAURORA-2Jを用いて評価実験を行った．その結果，特にMulticondition trainingの場合に提案法の有効性を認識することができた．","subitem_description_type":"Other"}]},"item_4_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"To achieve high recognition performance in noisy environments, we proposes the integration of recognition results obtained by using reduction algorithms (spectral subtraction with smoothing of time direction, temporal domain SVD-based speech enhancement, GMM-based speech estimation and KLT-based comb-filtering). In the proposed method, the log likelihood-based confidence measure is used to select the best recognition result. Experimental results on the AURORA-2J connected digit recognition task confirmed that the proposed algorithm achieves high recognition performance especially in the multicondition training.","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"114","bibliographic_titles":[{"bibliographic_title":"情報処理学会研究報告音声言語情報処理（SLP）"}],"bibliographicPageStart":"109","bibliographicIssueDates":{"bibliographicIssueDate":"2003-12-18","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"124(2003-SLP-049)","bibliographicVolumeNumber":"2003"}]},"relation_version_is_last":true,"weko_creator_id":"1"},"id":57169,"links":{}}