Acoustic features affecting speaker identification by imitated voice analysis

Mari Tanaka, Hideki Kawahara, Shigeo Morishima

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Abstract

    In this paper, physical correlates of perceived personal identity are investigated using imitated 16 utterances spoken by 11 mimicry speakers and 24 test subjects. Our unique strategy to use non-professional impersonators enabled to prepare test utterances with wide range of perceived similarities. Reasonably high correlations (0.46 and 0.44) in multiple regression analysis were attained by grouping subjects into three groups based on cluster analysis of the subjective test results. Without clustering, the correlation was only 0.17. Cluster analysis also revealed differences in their focusing physical correlates between three groups indicating importance of individual differences both in speakers and listeners.

    Original languageEnglish
    Title of host publication20th International Congress on Acoustics 2010, ICA 2010 - Incorporating Proceedings of the 2010 Annual Conference of the Australian Acoustical Society
    Pages3677-3680
    Number of pages4
    Volume5
    Publication statusPublished - 2010
    Event20th International Congress on Acoustics 2010, ICA 2010 - Incorporating the 2010 Annual Conference of the Australian Acoustical Society - Sydney, NSW
    Duration: 2010 Aug 232010 Aug 27

    Other

    Other20th International Congress on Acoustics 2010, ICA 2010 - Incorporating the 2010 Annual Conference of the Australian Acoustical Society
    CitySydney, NSW
    Period10/8/2310/8/27

    Fingerprint

    cluster analysis
    acoustics
    regression analysis

    ASJC Scopus subject areas

    • Acoustics and Ultrasonics

    Cite this

    Tanaka, M., Kawahara, H., & Morishima, S. (2010). Acoustic features affecting speaker identification by imitated voice analysis. In 20th International Congress on Acoustics 2010, ICA 2010 - Incorporating Proceedings of the 2010 Annual Conference of the Australian Acoustical Society (Vol. 5, pp. 3677-3680)

    Acoustic features affecting speaker identification by imitated voice analysis. / Tanaka, Mari; Kawahara, Hideki; Morishima, Shigeo.

    20th International Congress on Acoustics 2010, ICA 2010 - Incorporating Proceedings of the 2010 Annual Conference of the Australian Acoustical Society. Vol. 5 2010. p. 3677-3680.

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

    Tanaka, M, Kawahara, H & Morishima, S 2010, Acoustic features affecting speaker identification by imitated voice analysis. in 20th International Congress on Acoustics 2010, ICA 2010 - Incorporating Proceedings of the 2010 Annual Conference of the Australian Acoustical Society. vol. 5, pp. 3677-3680, 20th International Congress on Acoustics 2010, ICA 2010 - Incorporating the 2010 Annual Conference of the Australian Acoustical Society, Sydney, NSW, 10/8/23.
    Tanaka M, Kawahara H, Morishima S. Acoustic features affecting speaker identification by imitated voice analysis. In 20th International Congress on Acoustics 2010, ICA 2010 - Incorporating Proceedings of the 2010 Annual Conference of the Australian Acoustical Society. Vol. 5. 2010. p. 3677-3680
    Tanaka, Mari ; Kawahara, Hideki ; Morishima, Shigeo. / Acoustic features affecting speaker identification by imitated voice analysis. 20th International Congress on Acoustics 2010, ICA 2010 - Incorporating Proceedings of the 2010 Annual Conference of the Australian Acoustical Society. Vol. 5 2010. pp. 3677-3680
    @inproceedings{f5ccd5f31a764f69846638db4b58aa47,
    title = "Acoustic features affecting speaker identification by imitated voice analysis",
    abstract = "In this paper, physical correlates of perceived personal identity are investigated using imitated 16 utterances spoken by 11 mimicry speakers and 24 test subjects. Our unique strategy to use non-professional impersonators enabled to prepare test utterances with wide range of perceived similarities. Reasonably high correlations (0.46 and 0.44) in multiple regression analysis were attained by grouping subjects into three groups based on cluster analysis of the subjective test results. Without clustering, the correlation was only 0.17. Cluster analysis also revealed differences in their focusing physical correlates between three groups indicating importance of individual differences both in speakers and listeners.",
    author = "Mari Tanaka and Hideki Kawahara and Shigeo Morishima",
    year = "2010",
    language = "English",
    isbn = "9781617827457",
    volume = "5",
    pages = "3677--3680",
    booktitle = "20th International Congress on Acoustics 2010, ICA 2010 - Incorporating Proceedings of the 2010 Annual Conference of the Australian Acoustical Society",

    }

    TY - GEN

    T1 - Acoustic features affecting speaker identification by imitated voice analysis

    AU - Tanaka, Mari

    AU - Kawahara, Hideki

    AU - Morishima, Shigeo

    PY - 2010

    Y1 - 2010

    N2 - In this paper, physical correlates of perceived personal identity are investigated using imitated 16 utterances spoken by 11 mimicry speakers and 24 test subjects. Our unique strategy to use non-professional impersonators enabled to prepare test utterances with wide range of perceived similarities. Reasonably high correlations (0.46 and 0.44) in multiple regression analysis were attained by grouping subjects into three groups based on cluster analysis of the subjective test results. Without clustering, the correlation was only 0.17. Cluster analysis also revealed differences in their focusing physical correlates between three groups indicating importance of individual differences both in speakers and listeners.

    AB - In this paper, physical correlates of perceived personal identity are investigated using imitated 16 utterances spoken by 11 mimicry speakers and 24 test subjects. Our unique strategy to use non-professional impersonators enabled to prepare test utterances with wide range of perceived similarities. Reasonably high correlations (0.46 and 0.44) in multiple regression analysis were attained by grouping subjects into three groups based on cluster analysis of the subjective test results. Without clustering, the correlation was only 0.17. Cluster analysis also revealed differences in their focusing physical correlates between three groups indicating importance of individual differences both in speakers and listeners.

    UR - http://www.scopus.com/inward/record.url?scp=84869122991&partnerID=8YFLogxK

    UR - http://www.scopus.com/inward/citedby.url?scp=84869122991&partnerID=8YFLogxK

    M3 - Conference contribution

    AN - SCOPUS:84869122991

    SN - 9781617827457

    VL - 5

    SP - 3677

    EP - 3680

    BT - 20th International Congress on Acoustics 2010, ICA 2010 - Incorporating Proceedings of the 2010 Annual Conference of the Australian Acoustical Society

    ER -