The defender's perspective on automatic speaker verification: An overview
It tackles security vulnerabilities in ASV systems, which is critical for applications like authentication, but is incremental as it reviews existing defenses rather than proposing new solutions.
This paper addresses the problem of spoofing attacks, including adversarial attacks and partially fake speech, undermining automatic speaker verification (ASV) in security-sensitive environments by providing a comprehensive review of defense methods against these threats.
Automatic speaker verification (ASV) plays a critical role in security-sensitive environments. Regrettably, the reliability of ASV has been undermined by the emergence of spoofing attacks, such as replay and synthetic speech, as well as adversarial attacks and the relatively new partially fake speech. While there are several review papers that cover replay and synthetic speech, and adversarial attacks, there is a notable gap in a comprehensive review that addresses defense against adversarial attacks and the recently emerged partially fake speech. Thus, the aim of this paper is to provide a thorough and systematic overview of the defense methods used against these types of attacks.