Dear participants, for the final due diligence on your submission, please use the below README format to describe your submission. We also encourage you to add useful comments in your code to help streamline the review process.
-
Submission ID: XXXXX
-
Submitter: XXXXX
-
Final rank: Xth place on leaderboard X
-
Final scores on MDXDB21:
SDR_song SDR_bass SDR_drums SDR_other SDR_vocals X.XX X.XX X.XX X.XX X.XX
Please describe here your approach, including
- the model that you used (please cite relevant references, e.g., [1] if you use open-unmix)
- the training and validation dataset you used
- hyperparameters that you tuned (e.g., blending strength for system that blends two models)
[1] Stöter, Fabian-Robert, et al. "Open-unmix-a reference implementation for music source separation." Journal of Open Source Software 4.41 (2019): 1667.
Please describe here how your submission can be reproduced.
Please describe here how your model could be trained.