Applying base value of fundamental frequency via the multivariate Kernel-Density in Forensic Speaker Comparison

Forensic Speaker Comparison (FSC) is used in a legal process to compare speech patterns from a known speaker, the suspect and from an unknown offender, producing evidence for the conviction or the elimination of the suspect. The FSC methodology used most, applied worldwide by forensic experts, is based on linguistic-acoustic analysis. Fundamental Frequency (F0) is the acoustic parameter used most due to its robustness in poor quality audio recordings. The statistical base value parameter of F0 has shown, in recent research, to be more stable than other F0 measures and less affected by the speech style, the content, the recording channel and the speaker effort. Besides these good characteristics, only a small number of FSC experts worldwide use the base value parameter. In this research the discriminating power of the base value of F0 was investigated, as well as the improvement achieved by combining it to other statistical long-term measures of F0 using the Multivariate Kernel Density (MVKD). An Equal Error Rate (EER) of 13 % was obtained combining the statistical measure base value and the median of the F0, outperforming recent researches, indicating that the base value is more suitable in F0 acoustic analysis.

[1]  Anders Eriksson,et al.  Robustness of long time measures of fundamental frequency , 2007, INTERSPEECH.

[2]  Philip Rose,et al.  FORENSIC SPEAKER DISCRIMINATION WITH AUSTRALIAN ENGLISH VOWEL ACOUSTICS , 2007 .

[3]  Peter Dalgaard,et al.  R Development Core Team (2010): R: A language and environment for statistical computing , 2010 .

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

[5]  Pablo Arantes,et al.  Temporal stability of long-term measures of fundamental frequency , 2014 .

[6]  Philip Rose,et al.  Exploring the Discriminatory Potential of F0 Distribution Parameters in Traditional Forensic Speaker Recognition , 2009 .

[7]  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.

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

[9]  Geoffrey Stewart Morrison,et al.  A comparison of procedures for the calculation of forensic likelihood ratios from acoustic-phonetic data: Multivariate kernel density (MVKD) versus Gaussian mixture model-universal background model (GMM-UBM) , 2011, Speech Commun..

[10]  Anders Eriksson,et al.  The frequency range of the voice fundamental in the speech of male and female adults , 1993 .

[11]  Kurt Hornik,et al.  Misc Functions of the Department of Statistics, ProbabilityTheory Group (Formerly: E1071), TU Wien , 2015 .

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

[13]  H. Traunmüller.  Conventional, Biological and Environmental Factors in Speech Communication: A Modulation Theory , 1994, Phonetica.

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

[15]  J. Costa,et al.  Aplicação do Valor de Base da Frequência Fundamental Via Estatística MVKD em Comparação Forense de Locutor , 2016 .

[16]  Yuko Kinoshita.  Does Lindley's LR estimation formula work for speech data? Investigation using long-term f0 , 2005 .

[17]  M. Wand.  Functions for Kernel Smoothing Supporting Wand & Jones (1995) , 2015 .

[18]  P. Boersma.  ACCURATE SHORT-TERM ANALYSIS OF THE FUNDAMENTAL FREQUENCY AND THE HARMONICS-TO-NOISE RATIO OF A SAMPLED SOUND , 1993 .

[19]  Erica Gold,et al.  Calculating likelihood ratios for forensic speaker comparisons using phonetic and linguistic parameters , 2014 .

[20]  H. Traunmüller,et al.  The perceptual evaluation of F0 excursions in speech as evidenced in liveliness estimations. , 1995, The Journal of the Acoustical Society of America.