Supervised Contrastive Learning from Weakly-Labeled Audio Segments for Musical Version Matching.

Joan Serr, Recep Oguz Araz, Dmitry Bogdanov, Yuki Mitsufuji

Browse the full ICML paper archive.