Research2h ago
Apple Research Improves Text-to-Sounding Video Sync
Apple researchers developed a new framework to synchronize audio and video generation from text prompts. The method solves modal interference caused by shared captions and bridges the gap between dense training data and short user inputs. This refinement improves cross-modal feature interaction. Practitioners can expect more coherent audio-visual alignment in synthetic video generation.