Laion Audio 630K Dec 06, 2023 permanent A dataset described in paper Large-scale Contrastive Language-Audio Pre-training with Feature Fusion and Keyword-to-Caption Augmentation.