{"metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00057467","sets":["1164:5159:5210:5211"]},"path":["5211"],"owner":"1","recid":"57467","title":["端点検出を行わない連続音声認識手法"],"pubdate":{"attribute_name":"公開日","attribute_value":"2000-12-21"},"_buckets":{"deposit":"c5b58754-adaa-47f8-bdaf-9049b7c7c03a"},"_deposit":{"id":"57467","pid":{"type":"depid","value":"57467","revision_id":0},"owners":[1],"status":"published","created_by":1},"item_title":"端点検出を行わない連続音声認識手法","author_link":["0","0"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"端点検出を行わない連続音声認識手法"},{"subitem_title":"Continuous Speech Recognition without End - point Detection","subitem_title_language":"en"}]},"item_type_id":"4","publish_date":"2000-12-21","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"名古屋大学大学院工学研究科／中部電力株式会社電力技術研究所"},{"subitem_text_value":"名古屋大学大学院工学研究科"},{"subitem_text_value":"名古屋大学情報メディア教育センター"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Graduate School of Engineering, Nagoya University/Chubu Electric Power Co., Inc. Electric Power R&D Center","subitem_text_language":"en"},{"subitem_text_value":"Graduate School of Engineering, Nagoya University","subitem_text_language":"en"},{"subitem_text_value":"Center for Information Media Studies, Nagoya University","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/57467/files/IPSJ-SLP00034018.pdf"},"date":[{"dateType":"Available","dateValue":"2002-12-21"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-SLP00034018.pdf","filesize":[{"value":"571.9 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":"3aeaffa2-69b3-45b3-8671-cfe2b96d374d","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2000 by the Information Processing Society of Japan"}]},"item_4_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"瀬川, 修"},{"creatorName":"武田, 一哉"},{"creatorName":"板倉, 文忠"}],"nameIdentifiers":[{}]}]},"item_4_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Osamu, Segawa","creatorNameLang":"en"},{"creatorName":"Kazuya, Takeda","creatorNameLang":"en"},{"creatorName":"Fumitada, Itakura","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":"入力音声の明示的な端点検出を必要としない新しい連続音声認識手法を提案する。本手法では数秒程度の一定時間長の処理ブロックを認識すると同時に終端で途切れた単語区間をバックトラックによって修復しながら連続的にデコードを続けるため、端点検出や発話単位の考慮無しに無限長の人力音声を認識することが可能である。基本的なアルゴリズムは次のとおりである。1)一定時間長の処理ブロックの認識を行う。2)処理ブロックの終端フレームに残った全ての単語終端ノードよりトレースバックによって一つの最尤パスにマージするフレームを探し、これを最適な単語境界フレームとする。3)単語境界フレームまで戻ってサーチを再開する。本稿ではまずアルゴリズムの詳細を説明し、提案手法の有効性を検証するために行った約10分の連続した新聞読み上げ音声および男女各1名による約30分の車内音声対話の自動書き起こし実験の結果を示す。","subitem_description_type":"Other"}]},"item_4_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"A new continuous speech recognition method that does not need the explicit speech end-point detection is proposed. In this method, the decoder proceeds to recognize a processing block of a predetermined length and then to fix a word section which is broken at the end of a processing block. Therefore, continuous speech recognition of infinite length can be executed without the explicit end-point detection and without considering an utterance unit. The basic algorithm is 1) decode a processing block of the predetermined length, 2) traceback and find the boundaries of the processing blocks where the word history in the preceding processing block is merged into one, and 3) restart decoding from the boundary frame with the merged word history. The effectiveness of the method is verified by the two dictating experiments.","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"106","bibliographic_titles":[{"bibliographic_title":"情報処理学会研究報告音声言語情報処理（SLP）"}],"bibliographicPageStart":"101","bibliographicIssueDates":{"bibliographicIssueDate":"2000-12-21","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"119(2000-SLP-034)","bibliographicVolumeNumber":"2000"}]},"relation_version_is_last":true,"weko_creator_id":"1"},"id":57467,"updated":"2025-01-22T04:29:09.285660+00:00","links":{},"created":"2025-01-18T23:20:39.870117+00:00"}