Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2603.10713 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910050263498752 |
|---|---|
| author | Kushnir, Evgeny Kozodaev, Alexandr Korzh, Dmitrii Pautov, Mikhail Kiriukhin, Oleg Rogov, Oleg Y. |
| author_facet | Kushnir, Evgeny Kozodaev, Alexandr Korzh, Dmitrii Pautov, Mikhail Kiriukhin, Oleg Rogov, Oleg Y. |
| contents | Recent advances in generative models have amplified the risk of malicious misuse of speech synthesis technologies, enabling adversaries to impersonate target speakers and access sensitive resources. Although speech deepfake detection has progressed rapidly, most existing countermeasures lack formal robustness guarantees or fail to generalize to unseen generation techniques. We propose PV-VASM, a probabilistic framework for verifying the robustness of voice anti-spoofing models (VASMs). PV-VASM estimates the probability of misclassification under text-to-speech (TTS), voice cloning (VC), and parametric signal transformations. The approach is model-agnostic and enables robustness verification against unseen speech synthesis techniques and input perturbations. We derive a theoretical upper bound on the error probability and validate the method across diverse experimental settings, demonstrating its effectiveness as a practical robustness verification tool. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_10713 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Probabilistic Verification of Voice Anti-Spoofing Models Kushnir, Evgeny Kozodaev, Alexandr Korzh, Dmitrii Pautov, Mikhail Kiriukhin, Oleg Rogov, Oleg Y. Sound Artificial Intelligence Recent advances in generative models have amplified the risk of malicious misuse of speech synthesis technologies, enabling adversaries to impersonate target speakers and access sensitive resources. Although speech deepfake detection has progressed rapidly, most existing countermeasures lack formal robustness guarantees or fail to generalize to unseen generation techniques. We propose PV-VASM, a probabilistic framework for verifying the robustness of voice anti-spoofing models (VASMs). PV-VASM estimates the probability of misclassification under text-to-speech (TTS), voice cloning (VC), and parametric signal transformations. The approach is model-agnostic and enables robustness verification against unseen speech synthesis techniques and input perturbations. We derive a theoretical upper bound on the error probability and validate the method across diverse experimental settings, demonstrating its effectiveness as a practical robustness verification tool. |
| title | Probabilistic Verification of Voice Anti-Spoofing Models |
| topic | Sound Artificial Intelligence |
| url | https://arxiv.org/abs/2603.10713 |