Speaker Impersonation on Speaker Recognition Systems (SRSs)
7,500Number of QueriesSirenAttack
Evaluation results
| Method | Links | |
|---|---|---|
| SirenAttackImpersonation Type=Adversarial Example, Target Model Access=Black-box, Victim's Voice Requirement=Required, Strategy=Adaptive Query, Query Output Format=Score2026.03 | 7,500 | |
| Ours-NESVictim Voice Required=No, Query Output=Score2026.03 | 500 | |
| Ours-NESImpersonation Type=Speech Synthesis, Target Model Access=Black-box, Victim's Voice Requirement=Not Required, Strategy=Adaptive Query, Query Output Format=Score2026.03 | 500 | |
| VoxstructorVictim Voice Required=Yes, Query Output=Feature vector2026.03 | 100 | |
| VoxstructorImpersonation Type=Speech Synthesis, Target Model Access=Black-box, Victim's Voice Requirement=Required, Strategy=Training-based, Query Output Format=Feature vector2026.03 | 100 | |
| Ours-SPVictim Voice Required=No, Query Output=Score2026.03 | 50 | |
| Ours-SPImpersonation Type=Speech Synthesis, Target Model Access=Black-box, Victim's Voice Requirement=Not Required, Strategy=Non-Adaptive Query, Query Output Format=Score2026.03 | 50 | |
| OccamVictim Voice Required=Yes, Query Output=Decision2026.03 | 10 | |
| FakeBobVictim Voice Required=No, Query Output=Score2026.03 | 10 | |
| OccamImpersonation Type=Adversarial Example, Target Model Access=Black-box, Victim's Voice Requirement=Required, Strategy=Adaptive Query, Query Output Format=Decision2026.03 | 10 | |
| FakeBob†Impersonation Type=Adversarial Example, Target Model Access=Black-box, Victim's Voice Requirement=Not Required, Strategy=Adaptive Query, Query Output Format=Score2026.03 | 10 | |
| SirenAttackVictim Voice Required=Yes, Query Output=Score2026.03 | 7.5 | |
| [37]Victim Voice Required=Yes, Query Output=–2026.03 | 0 | |
| [84]Victim Voice Required=Yes, Query Output=–2026.03 | 0 | |
| [58]Victim Voice Required=Yes, Query Output=–2026.03 | 0 | |
| SV2TTSVictim Voice Required=Yes, Query Output=–2026.03 | 0 | |
| AdvTTSVictim Voice Required=Yes, Query Output=–2026.03 | 0 | |
| QFA2SRVictim Voice Required=Yes, Query Output=–2026.03 | 0 | |
| Voice MimicryImpersonation Type=Voice Mimicry, Target Model Access=Black-box, Victim's Voice Requirement=Required, Strategy=Human Mimicry, Query Output Format=-2026.03 | 0 | |
| Replay AttackImpersonation Type=Replay Attack, Target Model Access=Black-box, Victim's Voice Requirement=Required, Strategy=Playback, Query Output Format=-2026.03 | 0 | |
| Voice MorphingImpersonation Type=Voice Morphing, Target Model Access=Black-box, Victim's Voice Requirement=Required, Strategy=Identity Morphing, Query Output Format=-2026.03 | 0 | |
| SV2TTSImpersonation Type=Speech Synthesis, Target Model Access=White-box, Victim's Voice Requirement=Required, Strategy=Training-based, Query Output Format=-2026.03 | 0 | |
| AdvTTSImpersonation Type=Adversarial Example, Target Model Access=Black-box, Victim's Voice Requirement=Required, Strategy=Transfer-based, Query Output Format=-2026.03 | 0 | |
| QFA2SRImpersonation Type=Adversarial Example, Target Model Access=Black-box, Victim's Voice Requirement=Required, Strategy=Transfer-based, Query Output Format=-2026.03 | 0 |