Strength of forensic speaker identification evidence: Multispeaker formant and cepstrum-based segmental discrimination with a Bayesian likelihood ratio as threshold

Phil Rose, Takashi Osanai, Yuko Kinoshita

    Research output: Contribution to journalArticle

    23 Citations (Scopus)

    Abstract

    A forensic-phonetic speaker identification experiment is described which tests to what extent same-speaker pairs from a 60 speaker Japanese data base can be discriminated from different-speaker pairs using a Bayesian likelihood ratio (LR) as discriminant function. Non-contemporaneous telephone recordings are used, with comparison based on mean values from three segments only: a nasal, a voiceless fricative, and a vowel. It is shown that discrimination using the LR-based distance is better than with a conventional distance, and that the cepstrum outperforms the formants. A LR for the test of 50 is obtained for formant-based discrimination, compared to c. 900 for the cepstrum, and the tests are thus shown to be capable of yielding a probative strength of support for the prosecution hypothesis that is conventionally quantified as ‘moderate’ for formants but ‘moderately strong’ for the cepstrum. Comparisons are made with results from similar experiments
    Original languageEnglish
    Pages (from-to)179-202
    Number of pages24
    JournalForensic Linguistics
    Volume10
    Issue number2
    DOIs
    Publication statusPublished - 2003

    Fingerprint

    discrimination
    evidence
    experiment
    prosecution
    phonetics
    telephone
    recording

    Cite this

    @article{3eb2cc5aba704ae8857b7029a6a97bd8,
    title = "Strength of forensic speaker identification evidence: Multispeaker formant and cepstrum-based segmental discrimination with a Bayesian likelihood ratio as threshold",
    abstract = "A forensic-phonetic speaker identification experiment is described which tests to what extent same-speaker pairs from a 60 speaker Japanese data base can be discriminated from different-speaker pairs using a Bayesian likelihood ratio (LR) as discriminant function. Non-contemporaneous telephone recordings are used, with comparison based on mean values from three segments only: a nasal, a voiceless fricative, and a vowel. It is shown that discrimination using the LR-based distance is better than with a conventional distance, and that the cepstrum outperforms the formants. A LR for the test of 50 is obtained for formant-based discrimination, compared to c. 900 for the cepstrum, and the tests are thus shown to be capable of yielding a probative strength of support for the prosecution hypothesis that is conventionally quantified as ‘moderate’ for formants but ‘moderately strong’ for the cepstrum. Comparisons are made with results from similar experiments",
    author = "Phil Rose and Takashi Osanai and Yuko Kinoshita",
    year = "2003",
    doi = "10.1558/sll.2003.10.2.179",
    language = "English",
    volume = "10",
    pages = "179--202",
    journal = "Forensic Linguistics",
    issn = "1350-1771",
    publisher = "Equinox Publishing Ltd",
    number = "2",

    }

    Strength of forensic speaker identification evidence: Multispeaker formant and cepstrum-based segmental discrimination with a Bayesian likelihood ratio as threshold. / Rose, Phil; Osanai, Takashi; Kinoshita, Yuko.

    In: Forensic Linguistics, Vol. 10, No. 2, 2003, p. 179-202.

    Research output: Contribution to journalArticle

    TY - JOUR

    T1 - Strength of forensic speaker identification evidence: Multispeaker formant and cepstrum-based segmental discrimination with a Bayesian likelihood ratio as threshold

    AU - Rose, Phil

    AU - Osanai, Takashi

    AU - Kinoshita, Yuko

    PY - 2003

    Y1 - 2003

    N2 - A forensic-phonetic speaker identification experiment is described which tests to what extent same-speaker pairs from a 60 speaker Japanese data base can be discriminated from different-speaker pairs using a Bayesian likelihood ratio (LR) as discriminant function. Non-contemporaneous telephone recordings are used, with comparison based on mean values from three segments only: a nasal, a voiceless fricative, and a vowel. It is shown that discrimination using the LR-based distance is better than with a conventional distance, and that the cepstrum outperforms the formants. A LR for the test of 50 is obtained for formant-based discrimination, compared to c. 900 for the cepstrum, and the tests are thus shown to be capable of yielding a probative strength of support for the prosecution hypothesis that is conventionally quantified as ‘moderate’ for formants but ‘moderately strong’ for the cepstrum. Comparisons are made with results from similar experiments

    AB - A forensic-phonetic speaker identification experiment is described which tests to what extent same-speaker pairs from a 60 speaker Japanese data base can be discriminated from different-speaker pairs using a Bayesian likelihood ratio (LR) as discriminant function. Non-contemporaneous telephone recordings are used, with comparison based on mean values from three segments only: a nasal, a voiceless fricative, and a vowel. It is shown that discrimination using the LR-based distance is better than with a conventional distance, and that the cepstrum outperforms the formants. A LR for the test of 50 is obtained for formant-based discrimination, compared to c. 900 for the cepstrum, and the tests are thus shown to be capable of yielding a probative strength of support for the prosecution hypothesis that is conventionally quantified as ‘moderate’ for formants but ‘moderately strong’ for the cepstrum. Comparisons are made with results from similar experiments

    U2 - 10.1558/sll.2003.10.2.179

    DO - 10.1558/sll.2003.10.2.179

    M3 - Article

    VL - 10

    SP - 179

    EP - 202

    JO - Forensic Linguistics

    JF - Forensic Linguistics

    SN - 1350-1771

    IS - 2

    ER -