---
pretty_name: Music Demixing Benchmarks Synthetic
tags:
- audio
- audio-to-audio
- source-separation
- audio-source-separation
- music-source-separation
- music-demixing
- benchmark
- leaderboard
- objective-evaluation
- objective-metrics
- mvsep
- quality-checker
license: other
task_categories:
- audio-to-audio
size_categories:
- n<1K
---

# Music Demixing Benchmarks Synthetic

Unofficial, unpacked mirror of MVSep Quality Checker [Synthetic dataset](https://mvsep.com/quality_checker).

- [Synth Leaderboard (Full)](https://mvsep.com/quality_checker/synth_leaderboard)
- Official zip file: [1.9 GB](https://mvsep.com/storage/public/synth_dataset.zip)

## Download

**`hf` CLI and the `huggingface_hub` Python package are orders of magnitude faster than `git`.**

- Use
  - Hugging Face [Command Line Interface (CLI)](https://huggingface.co/docs/huggingface_hub/guides/cli)
  - git
  - `huggingface_hub` [Python package](https://huggingface.co/docs/huggingface_hub/installation)
- Choose
  - individual files
  - all files in the repository
  - subsets of files with filename pattern matching ([fnmatch](https://docs.python.org/3/library/fnmatch.html))
- Read more
  - Hugging Face [download guide](https://huggingface.co/docs/huggingface_hub/guides/download)
  - Hugging Face [repository and Git guide](https://huggingface.co/docs/hub/repositories-getting-started)

## Remember to use `--repo-type dataset` or `repo_type='dataset'`

## Example

```python
filenames_mixture: list[str] = ...

submissionDatasetType: str = ...

pathRoot: str = huggingface_hub.snapshot_download(f'MusicDemixingBenchmarks/{submissionDatasetType}'
                    , repo_type = 'dataset'
                    , allow_patterns = filenames_mixture
                )

pathFilenames_mixture: list[str] = list(map(partial(fs.path.combine, pathRoot), filenames_mixture))
```

## Genesis

```bibtex
@misc{solovyev2024benchmarksleaderboardssounddemixing,
      title={Benchmarks and leaderboards for sound demixing tasks},
      author={Roman Solovyev and Alexander Stempkovskiy and Tatiana Habruseva},
      year={2024},
      eprint={2305.07489},
      archivePrefix={arXiv},
      primaryClass={cs.SD},
      url={https://arxiv.org/abs/2305.07489},
}
```