{"id":169882,"metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00169882","sets":["1164:5159:8497:8861"]},"path":["8861"],"owner":"11","recid":"169882","title":["音声クエリによる音声検索語検出のための認識結果およびDNNベースの特徴抽出と再照合手法の比較評価"],"pubdate":{"attribute_name":"公開日","attribute_value":"2016-07-21"},"_buckets":{"deposit":"39926fcf-3dac-464c-9856-eb4eb0ba4ad7"},"_deposit":{"id":"169882","pid":{"type":"depid","value":"169882","revision_id":0},"owners":[11],"status":"published","created_by":11},"item_title":"音声クエリによる音声検索語検出のための認識結果およびDNNベースの特徴抽出と再照合手法の比較評価","author_link":["341902","341901","341904","341903","341905","341906"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"音声クエリによる音声検索語検出のための認識結果およびDNNベースの特徴抽出と再照合手法の比較評価"},{"subitem_title":"Rescoring with ASR output-based and DNN-based features extraction for improved query-by-example spoken term detection","subitem_title_language":"en"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"SPオーガナイズドセッション","subitem_subject_scheme":"Other"}]},"item_type_id":"4","publish_date":"2016-07-21","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"静岡大学大学院総合科学技術研究科"},{"subitem_text_value":"静岡大学大学院総合科学技術研究科"},{"subitem_text_value":"静岡大学大学院総合科学技術研究科"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Graduate School of Integrated Science and Technology, Shizuoka University","subitem_text_language":"en"},{"subitem_text_value":"Graduate School of Integrated Science and Technology, Shizuoka University","subitem_text_language":"en"},{"subitem_text_value":"Graduate School of Integrated Science and Technology, Shizuoka 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/169882/files/IPSJ-SLP16112011.pdf","label":"IPSJ-SLP16112011.pdf"},"date":[{"dateType":"Available","dateValue":"2018-07-21"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-SLP16112011.pdf","filesize":[{"value":"734.6 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":"3a90dbea-b1ba-4d13-9fd8-1d4cfeb5f82f","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2016 by the Information Processing Society of Japan"}]},"item_4_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"大石, 修司"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"松葉, 達也"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"甲斐, 充彦"}],"nameIdentifiers":[{}]}]},"item_4_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Shuji, Oishi","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Tatsuya, Matsuba","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Atsuhiko, Kai","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_source_id_11":{"attribute_name":"ISSN","attribute_value_mlt":[{"subitem_source_identifier":"2188-8663","subitem_source_identifier_type":"ISSN"}]},"item_4_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"近年，音声ドキュメント検索技術に関連した研究として，与えられた検索語が発話されている箇所を音声ドキュメント中から特定する音声検索語検出 (Spoken Term Detection：STD) の研究が盛んに行われている．本稿では，音声によるクエリ入力を想定した STD の手法を対象として考える．従来の STD 手法は，音声認識を行い検索対象及び検索語 (クエリ) を元にサブワード (音素や音節) 列などの認識結果を用いて検索を行う．以前に我々はサブワード単位音響モデルのパラメータから求める様々な音響的類似度を用いる方法を提案し，STD で検索性能を改善した．しかし音声クエリを用いる場合，未知語 (OOV) や誤認識の影響はより大きくなり，検索性能を低下させてしまう．そのため本稿では音声認識結果を用いる従来の STD 手法によるスポッティングを行った後，話者や環境の違いに頑健な DNN に基づいた特徴量によって再照合を行う手法を提案する．さらに，認識結果より得られる信頼度やクエリの長さの特徴を素性として書き起こしを含む開発用データによって自動的に構築する検出事例から学習したスコアリングモデルにより，正規化されたスコアを得る．これらの方法の併用により，更なる STD 精度の改善が得られた．","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"6","bibliographic_titles":[{"bibliographic_title":"研究報告音声言語情報処理（SLP）"}],"bibliographicPageStart":"1","bibliographicIssueDates":{"bibliographicIssueDate":"2016-07-21","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"11","bibliographicVolumeNumber":"2016-SLP-112"}]},"relation_version_is_last":true,"weko_creator_id":"11"},"updated":"2025-01-20T08:49:19.038265+00:00","created":"2025-01-19T00:40:36.674012+00:00","links":{}}