What Counts as Real? Speech Restoration and Voice Quality Conversion Pose New Challenges to Deepfake Detection
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866911515769044992 |
|---|---|
| author | Satish, Shree Harsha Bokkahalli Lameris, Harm Gustafson, Joakim Székely, Éva |
| author_facet | Satish, Shree Harsha Bokkahalli Lameris, Harm Gustafson, Joakim Székely, Éva |
| contents | Audio anti-spoofing systems are typically formulated as binary classifiers distinguishing bona fide from spoofed speech. This assumption fails under layered generative processing, where benign transformations introduce distributional shifts that are misclassified as spoofing. We show that phonation-modifying voice conversion and speech restoration are treated as out-of-distribution despite preserving speaker authenticity. Using a multi-class setup separating bona fide, converted, spoofed, and converted-spoofed speech, we analyse model behaviour through self-supervised learning (SSL) embeddings and acoustic correlates. The benign transformations induce a drift in the SSL space, compressing bona fide and spoofed speech and reducing classifier separability. Reformulating anti-spoofing as a multi-class problem improves robustness to benign shifts while preserving spoof detection, suggesting binary systems model the distribution of raw speech rather than authenticity itself. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_14033 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | What Counts as Real? Speech Restoration and Voice Quality Conversion Pose New Challenges to Deepfake Detection Satish, Shree Harsha Bokkahalli Lameris, Harm Gustafson, Joakim Székely, Éva Sound Artificial Intelligence Machine Learning Audio and Speech Processing Audio anti-spoofing systems are typically formulated as binary classifiers distinguishing bona fide from spoofed speech. This assumption fails under layered generative processing, where benign transformations introduce distributional shifts that are misclassified as spoofing. We show that phonation-modifying voice conversion and speech restoration are treated as out-of-distribution despite preserving speaker authenticity. Using a multi-class setup separating bona fide, converted, spoofed, and converted-spoofed speech, we analyse model behaviour through self-supervised learning (SSL) embeddings and acoustic correlates. The benign transformations induce a drift in the SSL space, compressing bona fide and spoofed speech and reducing classifier separability. Reformulating anti-spoofing as a multi-class problem improves robustness to benign shifts while preserving spoof detection, suggesting binary systems model the distribution of raw speech rather than authenticity itself. |
| title | What Counts as Real? Speech Restoration and Voice Quality Conversion Pose New Challenges to Deepfake Detection |
| topic | Sound Artificial Intelligence Machine Learning Audio and Speech Processing |
| url | https://arxiv.org/abs/2603.14033 |