From Talking to Singing: A New Challenge for Audio-Visual Deepfake Detection

With rapid advances in audio-visual generative models, reliable forgery detection becomes increasingly critical. Existing methods for audiovisual deepfake detection typically rely on crossmodal inconsistencies. In singing, rhythmic vocalization weakens this coupling and introduces a nontrivial domain shift, substantially degrading detection performance. We construct the Singing Head DeepFake (SHDF) dataset using rhythm-aware generative models to fill the gap in singing benchmarks. To cope with crossscenario domain shifts, we propose a Text-guided Audio-Visual Forgery Detection (T-AVFD) framework that generalizes across both talking and singing scenarios. T-AVFD comprises a facial authenticity pattern learner and a multi-modal differential weight learning module. The pattern learner aligns facial features with multigranularity textual descriptions to learn generalizable authenticity patterns. The weight learning module preserves intrinsic audio-visual consistency and adaptively integrates it with authenticity patterns via differential weighting. Extensive experiments on multiple talking head deepfake datasets and SHDF show consistent improvements over existing baselines and strong robustness under diverse perturbations. The project page is available at https://LiuKe3068LikWix. github.io/SingingHead-DeepFake/ .