Abstract:Supervised contrastive learning (SupCon) is widely used to shape representations, but has seen limited targeted study for audio deepfake detection. Existing work typically combines contrastive terms with broader pipelines; however, the focus on SupCon itself is missing. In this work, we run a controlled study on wav2vec2 XLS-R (300M) that varies (i) similarity in SupCon (cosine vs angular similarity derived from the hyperspherical angle) and (ii) negative scaling using a warm-started global cross-batch queue. Stage 1 fine-tunes the encoder and projection head with SupCon; Stage 2 freezes them and trains a linear classifier with BCE. Trained on ASVspoof 2019 LA and evaluated on ASV19 eval plus ITW and ASVspoof 2021 DF/LA, Cosine SupCon with a delayed queue achieves the best ITW EER (8.29%) and pooled EER (4.44), while angular similarity performs strongly without queued negatives (ITW 8.70), indicating reduced reliance on large negative sets.
| Subjects: | Audio and Speech Processing (eess.AS); Machine Learning (cs.LG) |
| Cite as: | arXiv:2604.26057 [eess.AS] |
| (or arXiv:2604.26057v1 [eess.AS] for this version) | |
| https://doi.org/10.48550/arXiv.2604.26057 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Hafiz Malik [view email]
[v1]
Tue, 28 Apr 2026 18:52:38 UTC (504 KB)