Proposes new cross-manipulation evaluation protocols for FakeAVCeleb and DeepSpeak v1, shows temporal jittering mitigates a leading-silence shortcut, and introduces the SIMBA baseline.
Do you really mean that? content driven audio-visual deepfake dataset and multimodal method for temporal forgery localization
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.MM 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
DeepFake Doctor: Diagnosing and Treating Audio-Video Fake Detection
Proposes new cross-manipulation evaluation protocols for FakeAVCeleb and DeepSpeak v1, shows temporal jittering mitigates a leading-silence shortcut, and introduces the SIMBA baseline.