Datasets:
File size: 3,014 Bytes
9bd1d61 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 | # /// script
# requires-python = ">=3.10"
# dependencies = ["pandas>=2"]
# ///
"""Assemble the Hugging Face Hub copy of FreeTZAN (https://huggingface.co/datasets/gclef-cmu/freetzan).
The Hub repo mirrors this repository plus the audio (audio/<split>/<genre>/<track_id>.mp3, not in git),
and adds:
audio/<split>/metadata.csv that split's rows of metadata.csv, with `file_name` relative to the folder;
the Hub dataset viewer and datasets.load_dataset read the clips through these
README.md this README.md with a YAML dataset-card header that maps audio/train,
audio/valid, audio/test to the Hub splits train, validation, test
Writes into --out (a clone of the Hub repo), replacing audio/, scripts/ and resources/ there. Push with
git + git-lfs (mp3 files are LFS-tracked by the Hub's default .gitattributes).
Usage: uv run scripts/make_hf_dataset.py --out /path/to/hub/clone
"""
import argparse
import hashlib
import shutil
from pathlib import Path
import pandas as pd
REPO = Path(__file__).resolve().parent.parent
SPLITS = ['train', 'valid', 'test']
MIRROR = ['LICENSE', 'metadata.csv', 'checksums.sha1', 'scripts/build_freetzan.py', 'scripts/find_duplicates.py',
'scripts/extract_features.py', 'scripts/run_baselines.py', 'scripts/make_hf_dataset.py',
'resources/exclusions.csv', 'resources/duplicates.csv', 'resources/baselines.csv']
CARD = """---
pretty_name: FreeTZAN
license: other
license_name: per-track-creative-commons
license_link: LICENSE
task_categories:
- audio-classification
tags:
- music
- genre-classification
size_categories:
- 1K<n<10K
configs:
- config_name: default
data_files:
- split: train
path: audio/train/**
- split: validation
path: audio/valid/**
- split: test
path: audio/test/**
---
"""
def main():
ap = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter)
ap.add_argument('--out', required=True)
args = ap.parse_args()
out = Path(args.out)
md = pd.read_csv(REPO / 'metadata.csv')
assert set(md.split) == set(SPLITS)
for d in ('audio', 'data', 'scripts', 'resources'):
shutil.rmtree(out / d, ignore_errors=True)
for f in MIRROR:
(out / f).parent.mkdir(parents=True, exist_ok=True)
shutil.copy2(REPO / f, out / f)
for p, sha1 in zip(md.path, md.sha1):
b = (REPO / p).read_bytes()
assert hashlib.sha1(b).hexdigest() == sha1, p
(out / p).parent.mkdir(parents=True, exist_ok=True)
(out / p).write_bytes(b)
for split in SPLITS:
rows = md.split == split
d = md[rows].drop(columns='path')
d.insert(0, 'file_name', md.path[rows].str.removeprefix(f'audio/{split}/'))
d.to_csv(out / 'audio' / split / 'metadata.csv', index=False)
print(f'{split}: {len(d)} clips')
(out / 'README.md').write_text(CARD + (REPO / 'README.md').read_text())
if __name__ == '__main__':
main()
|