@article{oai:u-fukui.repo.nii.ac.jp:02000190,
 author = {平塚, 喬 and 小高 , 知宏 and 黒岩, 丈介 and 白井, 治彦 and 諏訪, いずみ and Hiratuka , Kyou and Odaka, Tomohiro and Kuroiwa, Jousuke and Shirai, Haruhiko and Suwa, Izumi},
 issue = {12月},
 journal = {福井大学学術研究院工学系部門研究報告, Memoir of Faculty of Engineering, University of Fukui},
 month = {Mar},
 note = {In this paper, we tried to improve the authentication performance by increasing the training data of the speaker authentication model using voice conversion technology. We used one of the deep learning speaker authentication models, ”x-vector”, and increased the amount of data by incorporating data created by statistical voice conversion techniques as new speaker speech data when training the model. From the experiment, a comparison between models that incorporated data created by the voice conversion technology and those that did not, confirmed that models that incorporated data created by the voice conversion technology performed better, albeit slightly. This suggests that increasing the amount of data using voice conversion techniques is effective in learning speaker authentication models, but its impact was limited; therefore, more detailed study of data generation by voice conversion is needed in the future.},
 pages = {9--15},
 title = {声質変換技術によるデータ拡張を利用した話者認証モデルの学習},
 year = {2024},
 yomi = {ヒラツカ, キョウ and オダカ, トモヒロ and クロイワ, ジョウスケ and シライ, ハルヒコ and スワ, イズミ}
}