Research
52d agoApple Research Improves Text-to-Sounding Video Sync
A new study from Apple Machine Learning Research targets the synchronization gap in Text-to-Sounding-Video generation. Researchers introduced advanced modality conditioning to stop modal interference caused by shared captions. This approach better aligns concise user prompts with dense training data.
The Signal
It provides a more stable framework for generating audio-visual content from text.