Supervised contrastive learning from weakly-labeled audio segments for musical version matching

Open in new window