Metraipolis › archive › dc_cfac2697dd74

Not All Attacks Are Learned Equally in Speech Deepfake Detection — arXiv:2609.11763

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.11763
License (as archived)
CC0-1.0
Archived
2026-10-16T07:17:42.145Z
SHA-256 of the archived text
9c3d5d9e90b19bd5821517dddd80a9edb954dbbaa497c48bd61e64cf16449588
Stable URL of this page
https://metraipolis.com/documento/dc_cfac2697dd74

How to cite

Not All Attacks Are Learned Equally in Speech Deepfake Detection — arXiv:2609.11763 — arXiv (en), https://arxiv.org/abs/2609.11763 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_cfac2697dd74 (Metraipolis archive dc_cfac2697dd74, sha256 9c3d5d9e90b19bd5821517dddd80a9edb954dbbaa497c48bd61e64cf16449588)
BibTeX
@misc{dccfac2697dd74,
  title = {Not All Attacks Are Learned Equally in Speech Deepfake Detection — arXiv:2609.11763},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.11763},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_cfac2697dd74; sha256 9c3d5d9e90b19bd5821517dddd80a9edb954dbbaa497c48bd61e64cf16449588},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.11763. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Speech deepfake detection (SDD) models are trained on multi-attack datasets containing diverse spoofing systems, such as text-to-speech (TTS) and voice conversion (VC). In standard classifier training on multi-attack datasets, all attacks are treated as one spoofed class, and performance is reported using overall Equal Error Rate (EER). This aggregate view obscures how individual attacks shape learning and generalization. To better understand this attack-level behavior, we first balance TTS and VC exposure using sample and attack omission. We then measure attack-wise EER at inference and analyze attack-wise training loss and predictive entropy to characterize optimization. Results show that attacks contribute unequally: some attacks have high EER sensitivity and concentrated entropy with low loss, indicating strong influence on the decision boundary. We define these as high-impact attacks. To reduce uneven generalization across attacks, we propose a replay-regularized, attack-aware curriculum that steps exposure based on measured attack influence. Experiments on ASVspoof 2019, 2021, ASVspoof 5, and Fake-or-Real show improved overall robustness and reduced attack-level imbalance compared with standard multi-attack training.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-11 in eess.AS).
Categorie arXiv: Audio and Speech Processing (eess.AS).
Pagina del lavoro: https://arxiv.org/abs/2609.11763 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-16T06:59:25.295Z dalla pagina https://arxiv.org/catchup/eess.AS/2026-09-11?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).