{"metadata":{"_oai":{"id":"oai:ipsj.ixsq.nii.ac.jp:00142051","sets":["1164:5064:7900:8256"]},"path":["8256"],"owner":"11","recid":"142051","title":["Unsupervised pronunciation disambiguation of language model training corpora"],"pubdate":{"attribute_name":"公開日","attribute_value":"2015-05-16"},"_buckets":{"deposit":"78cbb975-489f-4aee-a5b1-878f0a096cc2"},"_deposit":{"id":"142051","pid":{"type":"depid","value":"142051","revision_id":0},"owners":[11],"status":"published","created_by":11},"item_title":"Unsupervised pronunciation disambiguation of language model training corpora","author_link":["209376","209375","209372","209378","209379","209380","209374","209373","209377","209381","209370","209371"],"item_titles":{"attribute_name":"タイトル","attribute_value_mlt":[{"subitem_title":"Unsupervised pronunciation disambiguation of language model training corpora"},{"subitem_title":"Unsupervised pronunciation disambiguation of language model training corpora","subitem_title_language":"en"}]},"item_type_id":"4","publish_date":"2015-05-16","item_4_text_3":{"attribute_name":"著者所属","attribute_value_mlt":[{"subitem_text_value":"IBM Research - Tokyo"},{"subitem_text_value":"IBM Research - Tokyo"},{"subitem_text_value":"IBM Research - Tokyo"},{"subitem_text_value":"IBM Research - Tokyo/Presently with Shizuoka University"},{"subitem_text_value":"Nuance Communications Inc."},{"subitem_text_value":"Nuance Communications Inc."}]},"item_4_text_4":{"attribute_name":"著者所属(英)","attribute_value_mlt":[{"subitem_text_value":"IBM Research - Tokyo","subitem_text_language":"en"},{"subitem_text_value":"IBM Research - Tokyo","subitem_text_language":"en"},{"subitem_text_value":"IBM Research - Tokyo","subitem_text_language":"en"},{"subitem_text_value":"IBM Research - Tokyo / Presently with Shizuoka University","subitem_text_language":"en"},{"subitem_text_value":"Nuance Communications Inc.","subitem_text_language":"en"},{"subitem_text_value":"Nuance Communications Inc.","subitem_text_language":"en"}]},"item_language":{"attribute_name":"言語","attribute_value_mlt":[{"subitem_language":"eng"}]},"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/142051/files/IPSJ-MUS15107065.pdf"},"date":[{"dateType":"Available","dateValue":"2017-05-16"}],"format":"application/pdf","billing":["billing_file"],"filename":"IPSJ-MUS15107065.pdf","filesize":[{"value":"628.0 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":"21"},{"tax":["include_tax"],"price":"0","billingrole":"44"}],"accessrole":"open_date","version_id":"07e2546a-b692-4341-846f-447060332779","displaytype":"detail","licensetype":"license_note","license_note":"Copyright (c) 2015 by the Information Processing Society of Japan"}]},"item_4_creator_5":{"attribute_name":"著者名","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Ryuki, Tachibana"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Nobuyasu, Itoh"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Gakuto, Kurata"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Masafumi, Nishimura"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Nicola, Ueffing"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Daniel, Willett"}],"nameIdentifiers":[{}]}]},"item_4_creator_6":{"attribute_name":"著者名(英)","attribute_type":"creator","attribute_value_mlt":[{"creatorNames":[{"creatorName":"Ryuki, Tachibana","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Nobuyasu, Itoh","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Gakuto, Kurata","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Masafumi, Nishimura","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Nicola, Ueffing","creatorNameLang":"en"}],"nameIdentifiers":[{}]},{"creatorNames":[{"creatorName":"Daniel, Willett","creatorNameLang":"en"}],"nameIdentifiers":[{}]}]},"item_4_source_id_9":{"attribute_name":"書誌レコードID","attribute_value_mlt":[{"subitem_source_identifier":"AN10438388","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":"2188-8752","subitem_source_identifier_type":"ISSN"}]},"item_4_description_7":{"attribute_name":"論文抄録","attribute_value_mlt":[{"subitem_description":"It is known that pronunciation probability estimation by an LM can improve the recognition accuracy of an ASR system. However, because training of such an LM usually requires manual preparation of corpora with pronunciation information, which is very costly, it is still a standard approach in the ASR research field to assume the same probability for all of the possible pronunciations of each word. In this paper, we avoid the cost by training a context-dependent pronunciation model in an unsupervised manner based on the recognition results of a large amount of user speech data. With this model, we can disambiguate the pronunciations of the sentences in the LM corpus. We also combine the model with a TTS frontend module to compensate for its inaccuracies. We will present results on a Japanese LVCSR task with a gain of 3.9% CERR.","subitem_description_type":"Other"}]},"item_4_description_8":{"attribute_name":"論文抄録(英)","attribute_value_mlt":[{"subitem_description":"It is known that pronunciation probability estimation by an LM can improve the recognition accuracy of an ASR system. However, because training of such an LM usually requires manual preparation of corpora with pronunciation information, which is very costly, it is still a standard approach in the ASR research field to assume the same probability for all of the possible pronunciations of each word. In this paper, we avoid the cost by training a context-dependent pronunciation model in an unsupervised manner based on the recognition results of a large amount of user speech data. With this model, we can disambiguate the pronunciations of the sentences in the LM corpus. We also combine the model with a TTS frontend module to compensate for its inaccuracies. We will present results on a Japanese LVCSR task with a gain of 3.9% CERR.","subitem_description_type":"Other"}]},"item_4_biblio_info_10":{"attribute_name":"書誌情報","attribute_value_mlt":[{"bibliographicPageEnd":"4","bibliographic_titles":[{"bibliographic_title":"研究報告音楽情報科学(MUS)"}],"bibliographicPageStart":"1","bibliographicIssueDates":{"bibliographicIssueDate":"2015-05-16","bibliographicIssueDateType":"Issued"},"bibliographicIssueNumber":"65","bibliographicVolumeNumber":"2015-MUS-107"}]},"relation_version_is_last":true,"weko_creator_id":"11"},"id":142051,"updated":"2025-01-20T19:09:50.227338+00:00","links":{},"created":"2025-01-19T00:19:33.190557+00:00"}