Abstract
The scope of this paper is purposefully limited to the 15 voice biometrics modalities discussed by Jain et al. (2004). The place of Voice within their classification scheme is reexamined in light of important developments that have taken place since 2010. Additionally, elements are added to Mayhew’s (2018) overview of the history of biometrics as an attempt to fill in missing gaps concerning Voice. All this leads to a reassessment of voice biometrics and how it relates to other biometric modalities. Speech segments that carry extremely high identity vector loads are discussed. The main assertion of this paper is that increased computing power, advanced algorithms, and the deployment of Artificial Intelligent have made voice biometrics optimal for use. Furthermore, the analysis of the compatibility among modalities, the estimation of inconvenience penalty, and the calculation of the arithmetic distances between various modalities indicate that the fusion of {Voice + Face}, {Voice + Fingerprint}, {Voice + Iris}, and {Voice + Signature} on the one hand, and of {Voice + Face +Fingerprint}, {Voice +Fingerprint + Signature} on the other, offer the best liveliness assurance against hacking, spoofing, and other malicious activities.
Recommended Citation
Koffi, Ettien
(2023)
"VOICE BIOMETRICS FUSION FOR ENHANCED SECURITY AND SPEAKER RECOGNITION: A COMPREHENSIVE REVIEW,"
Linguistic Portfolios: Vol. 12, Article 6.
Available at:
https://repository.stcloudstate.edu/stcloud_ling/vol12/iss1/6
