Velvet

Velvet is a data research lab focused exclusively on multimodal models. Its purpose is to help artificial intelligence succeed in audiovisual interpretation. Velvet provides tools and resources for research in audiovisual communication, enabling exploration of new worlds of interaction. The platform primarily targets researchers, engineers, and AI professionals aiming to better understand and improve multimodal communication. By offering a library of datasets and evaluations for audiovisual communication, Velvet seeks to advance research in multimodality.