Who Synthesized This? Joint Deepfake Detection and Generative Source Attribution
| Type of publication: | Conference paper |
| Citation: | Kumar_INTERSPEECH_2026 |
| Publication status: | Published |
| Booktitle: | Proceedings of Interspeech |
| Year: | 2026 |
| DOI: | 10.21437/Interspeech.2026-2442 |
| Abstract: | We propose a few-shot, open-set framework for joint deepfake detection and generative source attribution. Addressing the rapid evolution of synthetic speech, we introduce a temporal evaluation strategy using the MLAAD v9 dataset, partitioning models by launch year and architectural family. Our model utilizes a WavLM-Large model with targeted LoRA adapters, optimized via-a hierarchical metric learning objective: Conditional AAM Softmax + Center Loss. We simulate real-world forensic deployment by performing prototype-based inference on unseen 2025+ model families, establishing new class centroids with only 12 samples. Evaluated on the ASVspoof 5.0 benchmark, our framework achieves SOTA performance with 0.49% EER and 0.09 minDCF, while attaining 99% accuracy in tracing generated speech to its source. These results demonstrate that hierarchical representation learning and few-shot prototype adaptation provide a robust, scalable defense against the shifting landscape of audio synthesis. |
| Main Research Program: | Sustainable & Resilient Societies |
| Keywords: | Anti-spoofing, Open-World Recognition, Synthetic Speech Attribution |
| Projects: |
Idiap PaSS |
| Authors: | |
| Added by: | [UNK] |
| Total mark: | 0 |
|
Attachments
|
|
|
Notes
|
|
|
|
|