Regarding speaker identity in forensic conditions, several factors of variability must be taken into account, as peculiar intra-speaker variability, forced intra-speaker variability or channel-dependent external influences. Using 'AHUMADA' large speech database in Spanish, containing several recording sessions and channels, and including different tasks for 100 male speakers, automatic speaker verification experiments have been accomplished. Due to the inherent non-cooperative nature of speakers in forensic applications, only text-independent recognizers are used. In this sense, a GMM-based verification system is being used in order to obtain quantitative results. Maximum likelihood estimation of the models is performed, and LPC-cepstra, delta- and delta-delta-LPCC, are used at the parameterization stage. With this baseline verification system, we intend to determine how some variability sources included in 'AHUMADA' affect speaker identification. Results including speaking rate influence, single- and multi-session training, cross-channel testing, and kind of speech (read vs. spontaneous) are presented when likelihood-domain normalization is applied.
Cite as: Ortega-García, J., Cruz-Llanas, S., Gonzalez-Rodriguez, J. (1998) Quantitative influence of speech variability factors for automatic speaker verification in forensic tasks. Proc. 5th International Conference on Spoken Language Processing (ICSLP 1998), paper 1062, doi: 10.21437/ICSLP.1998-230
@inproceedings{ortegagarcia98_icslp,
author={Javier Ortega-García and Santiago Cruz-Llanas and Joaquin Gonzalez-Rodriguez},
title={{Quantitative influence of speech variability factors for automatic speaker verification in forensic tasks}},
year=1998,
booktitle={Proc. 5th International Conference on Spoken Language Processing (ICSLP 1998)},
pages={paper 1062},
doi={10.21437/ICSLP.1998-230}
}