Abstract
Automatic speaker verification (ASV) is a voice biometric technology whose security might be compromised by spoofing attacks. To increase the robustness against spoofing attacks, presentation attack detection (PAD) or anti-spoofing systems for detecting replay, text-to-speech and voice conversion-based spoofing attacks are being developed. However, it was recently shown that adversarial spoofing attacks may seriously fool anti-spoofing systems. Moreover, the robustness of the whole biometric system (ASV + PAD) against this new type of attack is completely unexplored. In this work, a new generative adversarial network for biometric anti-spoofing (GANBA) is proposed. GANBA has a twofold basis: (1) it jointly employs the anti-spoofing and ASV losses to yield very damaging adversarial spoofing attacks, and (2) it trains the PAD as a discriminator in order to make them more robust against these types of adversarial attacks. The proposed system is able to generate adversarial spoofing attacks which can fool the complete voice biometric system. Then, the resulting PAD discriminators of the proposed GANBA can be used as a defense technique for detecting both original and adversarial spoofing attacks. The physical access (PA) and logical access (LA) scenarios of the ASVspoof 2019 database were employed to carry out the experiments. The experimental results show that the GANBA attacks are quite effective, outperforming other adversarial techniques when applied in white-box and black-box attack setups. In addition, the resulting PAD discriminators are more robust against both original and adversarial spoofing attacks.
Highlights
Biometric authentication [1] aims to authenticate the identity claimed by a given individual based on the samples measured from biological characteristics
We propose a novel generative adversarial network for biometric anti-spoofing (GANBA) which generates adversarial spoofing attacks capable of fooling the presentation attack detection (PAD) system without being detected by the automatic speaker verification (ASV) system, i.e., without changing the speaker information of the utterance
We propose a generative adversarial network for biometric anti-spoofing (GANBA) in order to generate adversarial spoofing attacks and, at the same time, train the PAD discriminator in order to make it more robust against this type of attack
Summary
Biometric authentication [1] aims to authenticate the identity claimed by a given individual based on the samples measured from biological characteristics (e.g., voice, face, and fingerprints). Four types of spoofing attacks were identified by the scientific community [4]: (i) replay (i.e., using a pre-recorded voice of the target user), (ii) impersonation (i.e., mimicking the voice of the target voice), or either using (iii) text-to-speech synthesis (TTS) or (iv) voice conversion (VC) systems to generate artificial speech resembling the voice of a genuine user. These attacks can be presented to the ASV system using either logical access (LA) or physical access (PA) scenarios. There are three types of PLDA models [24]: simplified [25], standard [22], and two-covariance [26]
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have