Forensic speaker recognition in Chinese: A multivariate likelihood ratio discrimination on /i/ and /y/

Cuiling Zhang; Geoffrey Stewart Morrison; Philip Rose

Forensic speaker recognition in Chinese: A multivariate likelihood ratio discrimination on /i/ and /y/

Cuiling Zhang^*, Geoffrey Stewart Morrison, Philip Rose

^*Corresponding author for this work

School of Social Sciences and Humanities

Research output: Chapter in Book/Published conference output › Conference publication

Abstract

A likelihood-ratio-based forensic speaker discrimination was conducted using the mean formant frequencies of Standard Chinese /i/ and /y/ tokens produced by 64 male speakers. The speech data were relatively forensically realistic in that they were relatively extemporaneous, were recorded over the telephone, and were from three non-contemporaneous recording sessions. A multivariate-kernel-density formula was used to calculate cross-validated likelihood ratios comparing all possible same-speaker and different-speaker combinations across sessions. Results were comparable with those previously obtained with laboratory speech in other languages. In general, greater strength of evidence was obtained for recording sessions separated by one week than for recording sessions separated by one month.

Original language	English
Title of host publication	Proceedings of the Annual Conference of the International Speech Communication Association
Subtitle of host publication	INTERSPEECH 2008
Pages	1937-1940
Number of pages	4
Publication status	Published - 2008

Publication series

Name	Proceedings of Interspeech
ISSN (Print)	1990 9772

Keywords

Chinese
Forensic speaker recognition
Likelihood ratio

Cite this

@inproceedings{d70963f09a6d4a428c8378319554f488,

title = "Forensic speaker recognition in Chinese: A multivariate likelihood ratio discrimination on /i/ and /y/",

abstract = "A likelihood-ratio-based forensic speaker discrimination was conducted using the mean formant frequencies of Standard Chinese /i/ and /y/ tokens produced by 64 male speakers. The speech data were relatively forensically realistic in that they were relatively extemporaneous, were recorded over the telephone, and were from three non-contemporaneous recording sessions. A multivariate-kernel-density formula was used to calculate cross-validated likelihood ratios comparing all possible same-speaker and different-speaker combinations across sessions. Results were comparable with those previously obtained with laboratory speech in other languages. In general, greater strength of evidence was obtained for recording sessions separated by one week than for recording sessions separated by one month.",

keywords = "Chinese, Forensic speaker recognition, Likelihood ratio",

author = "Cuiling Zhang and Morrison, {Geoffrey Stewart} and Philip Rose",

year = "2008",

language = "English",

series = "Proceedings of Interspeech",

pages = "1937--1940",

booktitle = "Proceedings of the Annual Conference of the International Speech Communication Association",

}

Forensic speaker recognition in Chinese: A multivariate likelihood ratio discrimination on /i/ and /y/. / Zhang, Cuiling; Morrison, Geoffrey Stewart; Rose, Philip.
Proceedings of the Annual Conference of the International Speech Communication Association: INTERSPEECH 2008. 2008. p. 1937-1940 (Proceedings of Interspeech).

Research output: Chapter in Book/Published conference output › Conference publication

TY - GEN

T1 - Forensic speaker recognition in Chinese

T2 - A multivariate likelihood ratio discrimination on /i/ and /y/

AU - Zhang, Cuiling

AU - Morrison, Geoffrey Stewart

AU - Rose, Philip

PY - 2008

Y1 - 2008

N2 - A likelihood-ratio-based forensic speaker discrimination was conducted using the mean formant frequencies of Standard Chinese /i/ and /y/ tokens produced by 64 male speakers. The speech data were relatively forensically realistic in that they were relatively extemporaneous, were recorded over the telephone, and were from three non-contemporaneous recording sessions. A multivariate-kernel-density formula was used to calculate cross-validated likelihood ratios comparing all possible same-speaker and different-speaker combinations across sessions. Results were comparable with those previously obtained with laboratory speech in other languages. In general, greater strength of evidence was obtained for recording sessions separated by one week than for recording sessions separated by one month.

AB - A likelihood-ratio-based forensic speaker discrimination was conducted using the mean formant frequencies of Standard Chinese /i/ and /y/ tokens produced by 64 male speakers. The speech data were relatively forensically realistic in that they were relatively extemporaneous, were recorded over the telephone, and were from three non-contemporaneous recording sessions. A multivariate-kernel-density formula was used to calculate cross-validated likelihood ratios comparing all possible same-speaker and different-speaker combinations across sessions. Results were comparable with those previously obtained with laboratory speech in other languages. In general, greater strength of evidence was obtained for recording sessions separated by one week than for recording sessions separated by one month.

KW - Chinese

KW - Forensic speaker recognition

KW - Likelihood ratio

UR - http://www.scopus.com/inward/record.url?scp=84867197702&partnerID=8YFLogxK

UR - https://www.isca-speech.org/archive/interspeech_2008/i08_1937.html

M3 - Conference publication

AN - SCOPUS:84867197702

T3 - Proceedings of Interspeech

SP - 1937

EP - 1940

BT - Proceedings of the Annual Conference of the International Speech Communication Association

ER -

Forensic speaker recognition in Chinese: A multivariate likelihood ratio discrimination on /i/ and /y/

Abstract

Publication series

Keywords

Other files and links

Fingerprint

Cite this