{"created":"2025-01-19T01:31:27.509518+00:00","metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00231262","sets":["1164:4179:11237:11430"]},"path":["11430"],"owner":"44499","recid":"231262","title":["拡散モデルを用いた音声強調の計算量削減"],"pubdate":{"attribute_name":"公開日","attribute_value":"2023-11-25"},"_buckets":{"deposit":"c6636af7-0d2c-45b2-abe6-1f44c0c32b3c"},"_deposit":{"id":"231262","pid":{"type":"depid","value":"231262","revision_id":0},"owners":[44499],"status":"published","created_by":44499},"item_title":"拡散モデルを用いた音声強調の計算量削減","author_link":["624069","624067","624071","624068","624070","624066"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"拡散モデルを用いた音声強調の計算量削減"}]},"item_keyword":{"attribute_name":"キーワード","attribute_value_mlt":[{"subitem_subject":"モデルとデータ活用","subitem_subject_scheme":"Other"}]},"item_type_id":"4","publish_date":"2023-11-25","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"東京工業大学"},{"subitem_text_value":" 東京都市大学"},{"subitem_text_value":"東京工業大学"}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"Tokyo Insutitute of Technology","subitem_text_language":"en"},{"subitem_text_value":"Tokyo City University","subitem_text_language":"en"},{"subitem_text_value":"Tokyo Insutitute of Technology","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/231262/files/IPSJ-NL23258004.pdf","label":"IPSJ-NL23258004.pdf"},"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-NL23258004.pdf","filesize":[{"value":"980.0 kB"}],"mimetype":"application/pdf","priceinfo":[{"tax":["include_tax"],"price":"0","billingrole":"23"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_login","version_id":"4c341167-8e53-4891-ad0a-d03bbfcf14fb","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2023 by the Institute of Electronics, Information and Communication Engineers This SIG report is only available to those in membership of the SIG."}]},"item_4_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"西, 悠希"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"岩野, 公司"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"篠田, 浩一"}],"nameIdentifiers":[{}]}]},"item_4_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Yuki, Nishi","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Koji, Iwano","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Koichi, Shinoda","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_4_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AN10115061","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":"2188-8779","subitem_source_identifier_type":"ISSN"}]},"item_4_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"近年拡散モデルと呼ばれる生成モデルが注目されている．GAN と比べ，拡散モデルは安定に学習できるが，生成段階の計算コストが大きいという問題点がある．この傾向は音声強調への拡散モデルの応用に関しても同様である．本稿では，音声強調のための拡散モデルにおいて，Encoder，Decoder を用いることによる潜在空間にて音声信号を圧縮し，圧縮された信号から拡散モデルにより雑音を除去することで，精度を保ちつつ計算コストの削減することが可能なことを示す．雑音と音声を同時に用いる訓練で Encoder，Decoder を学習した結果，PESQ を低下させずに生成時間を 50% 以上減少させることに成功した．","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"6","bibliographic_titles":[{"bibliographic_title":"研究報告自然言語処理（NL）"}],"bibliographicPageStart":"1","bibliographicIssueDates":{"bibliographicIssueDate":"2023-11-25","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"4","bibliographicVolumeNumber":"2023-NL-258"}]},"relation_version_is_last":true,"weko_creator_id":"44499"},"links":{},"id":231262,"updated":"2025-01-19T10:49:55.044350+00:00"}