Can We Use Speaker Recognition Technology to Attack Itself? Enhancing Mimicry Attacks Using Automatic Target Speaker Selection

Tomi Kinnunen (1) , Rosa González Hautamäki (1) , Ville Vestman (1) , Md Sahidullah (2)

1 University of Eastern Finland
2 MULTISPEECH - Speech Modeling for Facilitating Oral-Based Communication

We consider technology-assisted mimicry attacks in the context of automatic speaker verification (ASV). We use ASV itself to select targeted speakers to be attacked by human-based mimicry. We recorded 6 naive mimics for whom we select target celebrities from VoxCeleb1 and VoxCeleb2 corpora (7,365 potential targets) using an i-vector system. The attacker attempts to mimic the selected target, with the utterances subjected to ASV tests using an independently developed x-vector system. Our main finding is negative: even if some of the attacker scores against the target speakers were slightly increased, our mimics did not succeed in spoofing the x-vector system. Interestingly, however, the relative ordering of the selected targets (closest, furthest, median) are consistent between the systems, which suggests some level of transferability between the systems

Mots clés: Speaker verification; mimicry; spoofing

Domaines: Intelligence artificielle [cs.AI]; Interface homme-machine [cs.HC]; Traitement du signal et de l'image [eess.SP]; Acoustique [physics.class-ph]

Tomi Kinnunen, Rosa González Hautamäki, Ville Vestman, Md Sahidullah. Can We Use Speaker Recognition Technology to Attack Itself? Enhancing Mimicry Attacks Using Automatic Target Speaker Selection. 2018. ⟨hal-01937767⟩