This project regards a new framework for deepfake detection, which pursues accuracy and explainability of detection.