[Invited Talk] Exploiting Redundant Coding of Message Information in Machine Recognition of Speech

• Item type: SIG Technical Reports(1)
• 公開日: 2017-12-14
• タイトル: [Invited Talk] Exploiting Redundant Coding of Message Information in Machine Recognition of Speech
• タイトル / 言語: en
• タイトル: [Invited Talk] Exploiting Redundant Coding of Message Information in Machine Recognition of Speech
• 言語: eng
• キーワード / 主題Scheme: Other
• キーワード / 主題: オーガナイズドセッション: 音声言語情報処理が切り拓く新しい情報社会
• 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: The Johns Hopkins University
• 著者所属(英): en
• 著者所属(英): The Johns Hopkins University
• 著者名: Hynek, Hermansky
• 著者名(英): Hynek, Hermansky
• 論文抄録 / 内容記述タイプ: Other
• 論文抄録 / 内容記述: In communicating the message, a speaker issues series of commands to create a percept of consecutive sounds of speech in a mind of a listener. The message needs to be coded in redundant way to survive noise during its transmission. Inertia of human vocal organs spreads the information about speech sounds redundantly in time and acoustics of a vocal tract redundantly distributes the information into different frequencies. Human hearing is capable of simultaneous processing of information within several centiseconds intervals and within individual articulation frequency bands, and uses the information spread in time and in frequency to recreate the message. The talk discusses possible ways of exploiting this concept in machine recognition of speech.
• 論文抄録(英) / 内容記述タイプ: Other
• 論文抄録(英) / 内容記述: In communicating the message, a speaker issues series of commands to create a percept of consecutive sounds of speech in a mind of a listener. The message needs to be coded in redundant way to survive noise during its transmission. Inertia of human vocal organs spreads the information about speech sounds redundantly in time and acoustics of a vocal tract redundantly distributes the information into different frequencies. Human hearing is capable of simultaneous processing of information within several centiseconds intervals and within individual articulation frequency bands, and uses the information spread in time and in frequency to recreate the message. The talk discusses possible ways of exploiting this concept in machine recognition of speech.
• 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: AN10442647
• 書誌情報: 研究報告音声言語情報処理（SLP）; 巻 2017-SLP-119, 号 13, p. 1-1, 発行日 2017-12-14
• ISSN / 収録物識別子タイプ: ISSN
• ISSN / 収録物識別子: 2188-8663
• 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