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

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Liu, Ke, Wei, Jiwei, Zhang, Wenyu, Zhou, Shuchang, Chai, Ruikun, Dai, Yutao, Zhang, Chaoning, Yang, Yang
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913166768734208
author Liu, Ke
Wei, Jiwei
Zhang, Wenyu
Zhou, Shuchang
Chai, Ruikun
Dai, Yutao
Zhang, Chaoning
Yang, Yang
author_facet Liu, Ke
Wei, Jiwei
Zhang, Wenyu
Zhou, Shuchang
Chai, Ruikun
Dai, Yutao
Zhang, Chaoning
Yang, Yang
contents With rapid advances in audio-visual generative models, reliable forgery detection becomes increasingly critical. Existing methods for audio-visual deepfake detection typically rely on cross-modal 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 cross-scenario 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 multi-granularity 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.
format Preprint
id arxiv_https___arxiv_org_abs_2605_27944
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle From Talking to Singing: A New Challenge for Audio-Visual Deepfake Detection
Liu, Ke
Wei, Jiwei
Zhang, Wenyu
Zhou, Shuchang
Chai, Ruikun
Dai, Yutao
Zhang, Chaoning
Yang, Yang
Artificial Intelligence
Multimedia
Sound
With rapid advances in audio-visual generative models, reliable forgery detection becomes increasingly critical. Existing methods for audio-visual deepfake detection typically rely on cross-modal 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 cross-scenario 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 multi-granularity 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.
title From Talking to Singing: A New Challenge for Audio-Visual Deepfake Detection
topic Artificial Intelligence
Multimedia
Sound
url https://arxiv.org/abs/2605.27944