CLEWS Checkpoints

This contains the checkpoints for the paper "Supervised Contrastive Learning from Weakly-Labeled Audio Segments for Musical Version Matching" by J. Serrà, R. O. Araz, D. Bogdanov & Y. Mitsufuji. 

Checkpoints are available for SHS and DVI data sets and for all models except ByteCover3x (due to size). For more information, please check the paper or the source code.

• Arxiv: https://arxiv.org/abs/2502.16936
• GitHub: https://github.com/sony/clews