Enhancing Multi-Accent Automated Speech Recognition with Accent-Activated Adapters

• Item type: SIG Technical Reports(1)
• 公開日: 2023-11-25
• タイトル: Enhancing Multi-Accent Automated Speech Recognition with Accent-Activated Adapters
• タイトル / 言語: en
• タイトル: Enhancing Multi-Accent Automated Speech Recognition with Accent-Activated Adapters
• 言語: eng
• キーワード / 主題Scheme: Other
• キーワード / 主題: ポスター
• 資源タイプ / 資源タイプ識別子: http://purl.org/coar/resource_type/c_18gh
• 資源タイプ: technical report
• 著者所属: Tianjin Key Laboratory of Cognitive Computing and Application, College of Intelligence and Computing, Tianjin University／Graduate School of Engineering, The University of Tokyo
• 著者所属: Tianjin Key Laboratory of Cognitive Computing and Application, College of Intelligence and Computing, Tianjin University
• 著者所属: Tianjin Key Laboratory of Cognitive Computing and Application, College of Intelligence and Computing, Tianjin University
• 著者所属: Graduate School of Engineering, The University of Tokyo
• 著者所属(英): en
• 著者所属(英): Tianjin Key Laboratory of Cognitive Computing and Application, College of Intelligence and Computing, Tianjin University / Graduate School of Engineering, The University of Tokyo
• 著者所属(英): en
• 著者所属(英): Tianjin Key Laboratory of Cognitive Computing and Application, College of Intelligence and Computing, Tianjin University
• 著者所属(英): en
• 著者所属(英): Tianjin Key Laboratory of Cognitive Computing and Application, College of Intelligence and Computing, Tianjin University
• 著者所属(英): en
• 著者所属(英): Graduate School of Engineering, The University of Tokyo
• 著者名: Yuqin, Lin; Longbiao, Wang; Jianwu, Dang; Nobuaki, Minematsu
• 著者名(英): Yuqin, Lin; Longbiao, Wang; Jianwu, Dang; Nobuaki, Minematsu
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: This paper proposes the Accent-Activated adapter (AccentAct) approach to address the challenge of speech variations in multi-accent scenarios. By incorporating parallel accent and contextual extractors within a pre-trained model, AccentAct improves ASR performance while reducing computational resources. Experimental results show that AccentAct outperforms traditional methods with a significant reduction in computational requirements, promoting inclusivity for individuals with diverse accents or dialects.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: This paper proposes the Accent-Activated adapter (AccentAct) approach to address the challenge of speech variations in multi-accent scenarios. By incorporating parallel accent and contextual extractors within a pre-trained model, AccentAct improves ASR performance while reducing computational resources. Experimental results show that AccentAct outperforms traditional methods with a significant reduction in computational requirements, promoting inclusivity for individuals with diverse accents or dialects.
• 書誌レコードID / 収録物識別子タイプ: NCID
• 書誌レコードID / 収録物識別子: AN10115061
• 書誌情報: 研究報告自然言語処理（NL）; 巻 2023-NL-258, 号 13, p. 1-6, 発行日 2023-11-25
• ISSN / 収録物識別子タイプ: ISSN
• ISSN / 収録物識別子: 2188-8779
• Notice: SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc.
• 出版者 / 言語: ja
• 出版者: 情報処理学会