Fetching the paper…

DiMoDif: Discourse Modality-information Differentiation for Audio-visual Deepfake Detection and Localization · Around