Version 1
: Received: 1 October 2024 / Approved: 1 October 2024 / Online: 1 October 2024 (13:02:16 CEST)
How to cite:
Fejes, A.; Saraiva, A.; Devenson, J. Assessing the impact of different face masks on results of Forensic Automatic Speaker Recognition systems. Preprints2024, 2024100067. https://doi.org/10.20944/preprints202410.0067.v1
Fejes, A.; Saraiva, A.; Devenson, J. Assessing the impact of different face masks on results of Forensic Automatic Speaker Recognition systems. Preprints 2024, 2024100067. https://doi.org/10.20944/preprints202410.0067.v1
Fejes, A.; Saraiva, A.; Devenson, J. Assessing the impact of different face masks on results of Forensic Automatic Speaker Recognition systems. Preprints2024, 2024100067. https://doi.org/10.20944/preprints202410.0067.v1
APA Style
Fejes, A., Saraiva, A., & Devenson, J. (2024). Assessing the impact of different face masks on results of Forensic Automatic Speaker Recognition systems. Preprints. https://doi.org/10.20944/preprints202410.0067.v1
Chicago/Turabian Style
Fejes, A., André Saraiva and Jelena Devenson. 2024 "Assessing the impact of different face masks on results of Forensic Automatic Speaker Recognition systems" Preprints. https://doi.org/10.20944/preprints202410.0067.v1
Abstract
Forensic speaker recognition plays a key role in criminal investigations, providing important conclusions for the justice system. The mandatory use of protection masks during the COVID-19 pandemic has posed a challenge for forensic speaker recognition, as they act as voice barriers or filters. Although the pandemic has been declared over, analyzing their impact in forensic speaker recognition contributes to a better understanding of the use of coverings as a voice disguise technique. This study aims to evaluate the impact of two types of face masks on an automatic forensic speech recognition system. For this purpose, the Multilingual Forensic Voice Database (FMVD), developed under the CERTAIN-FORS project, funded by the European Union, was used. Comparisons were made between dialog speech samples and reading samples collected without a mask, with a surgical mask, and with an FFP2 mask for both sexes in eight different languages. The performance metrics equal error rate (EER) and cost of likelihood ratio (Cllr) were calculated and analyzed. The results show that the presence of face masks has an impact on the performance metrics. The effect observed varies according to the language spoken, the gender of the speaker, and the type of mask.
Keywords
Forensic Automatic Speaker Recognition (FASR); Forensic Multilingual Voices Database (FMVD); face masks; COVID-19; system performance
Subject
Physical Sciences, Acoustics
Copyright:
This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.