Articulation Rate as a Speaker Discriminant in British English

Identifying speech parameters that have both a low level of intra-speaker variability and a high level of inter-speaker variability is key when discriminating between individuals in forensic speaker comparison cases. A substantial amount of research in the field of forensic phonetics has been devoted to identifying highly discriminant speaker parameters. To this end, the vast majority of the existing literature has focused solely on vowels and constants. However, the discriminant power of speaking tempo has yet to be examined, despite its broad use in practice and it having been recognized. This paper examines, for the first time, the discriminant power of articulation rate (AR) in British English. Approximately 3000 local ARs were measured in this study for 100 Southern Standard British English male speakers. In order to assess the evidential value of AR, likelihood ratios were calculated. The results suggest that AR performs well for same speaker comparisons. However, for different speaker comparisons, the system is performing just worse than chance. Overall, it appears that AR may not be the best speaker discriminant, although it is important to still consider AR in forensic speaker comparisons as there may be some individuals for which AR is highly idiosyncratic.

[1]  Michael Jessen,et al.  Forensic reference data on articulation rate in German. , 2007, Science & justice : journal of the Forensic Science Society.

[2]  David A. van Leeuwen,et al.  An Introduction to Application-Independent Evaluation of Speaker Recognition Systems , 2007, Speaker Classification.

[3]  G. Morrison.  Likelihood-ratio forensic voice comparison using parametric representations of the formant trajectories of diphthongs. , 2009, The Journal of the Acoustical Society of America.

[4]  John Laver,et al.  Principles of Phonetics: Principles of transcription , 1994 .

[5]  J. Trouvain,et al.  TEMPO VARIATION IN SPEECH PRODUCTION , 2003 .

[6]  Kirsty McDougall,et al.  Speaker-specific formant dynamics: An experiment on Australian English /aI/ , 2004 .

[7]  Colleen Kavanagh,et al.  New consonantal acoustic parameters for forensic speaker comparison , 2012 .

[8]  Vincent Hughes.  The definition of the relevant population and the collection of data for likelihood ratio-based forensic voice comparison , 2014 .

[9]  Niko Brümmer,et al.  Application-independent evaluation of speaker detection , 2006, Comput. Speech Lang..

[10]  Colin Aitken,et al.  Evaluation of trace evidence in the form of multivariate data , 2004 .

[11]  Hermann J. Künzel.  Some general phonetic and forensic aspects of speaking tempo , 2013 .

[12]  Philip Rose,et al.  The Intrinsic Forensic Discriminatory Power of Diphthongs , 2006 .

[13]  G. Morrison.  THE PLACE OF FORENSIC VOICE COMPARISON IN THE ONGOING PARADIGM SHIFT , 2009 .

[14]  Peter French,et al.  International practices in forensic speaker comparison , 2011 .

[15]  A. Stolcke,et al.  Speaker recognition using prosodic and lexical features , 2003, 2003 IEEE Workshop on Automatic Speech Recognition and Understanding (IEEE Cat. No.03EX721).

[16]  Wang Yingli,et al.  A Forensic Aspect of Articulation Rate Variation in Chinese , 2011 .

[17]  J. L. Miller,et al.  Articulation Rate and Its Variability in Spontaneous Speech: A Reanalysis and Some Implications , 1984, Phonetica.

[18]  Francis Nolan,et al.  The DyViS database: style-controlled recordings of 100 homogeneous speakers for forensic phonetic research , 2009 .