CoolFace
Datasetpublic

HughXuechen/fsd50k-cc0-curated-v1

FSD50K CC0 Curated v1 A 1,408-clip CC0-only subset of FSD50K (Fonseca et al., 2022), curated for an RNN/LSTM audio generation teaching assignment. Contents 1,408 WAV files from the FSD50K dev split (<file_id>.wav) fsd50k_cc0_dev_curated_v1_manifest.csv — per-clip metadata All files are CC0 / public domain — no attribution required 18 primary labels covering music instruments and nature ambient sounds Total size: ~1.5 GB, total duration: ~4.73 hours Original… See the full description on the dataset page: https://huggingface.co/datasets/HughXuechen/fsd50k-cc0-curated-v1.

sourceHugging Facecc0-1.0updated 5mo agoView on Hugging Face
0likes118downloads
Dataset Card

FSD50K CC0 Curated v1

A 1,408-clip CC0-only subset of FSD50K (Fonseca et al., 2022), curated for an RNN/LSTM audio generation teaching assignment.

Contents

  • —1,408 WAV files from the FSD50K dev split (<file_id>.wav)
  • —fsd50k_cc0_dev_curated_v1_manifest.csv — per-clip metadata
  • —All files are CC0 / public domain — no attribution required
  • —18 primary labels covering music instruments and nature ambient sounds
  • —Total size: ~1.5 GB, total duration: ~4.73 hours
  • —Original sample rates and durations preserved (no resampling or truncation)

Manifest columns

ColumnDescription
file_idFSD50K numeric ID — matches WAV filename
wav_pathOriginal local path (ignore — use file_id to locate files)
splitTrain / val / test assignment (stratified by primary_label, seed=42)
license_rawOriginal Freesound license URL
license_normalizedAlways CC0 in this subset
labelsComma-separated FSD50K labels
primary_labelSingle label used for stratification (first keep-label match)
tagsFreesound user tags
size_bytesWAV file size in bytes
duration_secClip duration in seconds

Curation

Clips were selected from the 14,959 CC0 dev files using:

  1. 1.Exclude-label filter (speech, vehicles, household noise, etc.)
  2. 2.Keep-label requirement (music + nature ambient categories)
  3. 3.Duration threshold ≥ 4.17 s (with 6 anchor clip exceptions)
  4. 4.Round-robin category-balanced fill to a 1.5 GB ceiling

Seed: 42. Full details in the source repository's doc/DATASET_CURATION_REPORT.md.

Usage

python
from huggingface_hub import snapshot_download
snapshot_download(repo_id="hughxuechen/fsd50k-cc0-curated-v1", repo_type="dataset", local_dir="data/fsd50k_preprocessed")

Citation

Fonseca, E., Favory, X., Pons, J., Font, F., & Serra, X. (2022). FSD50K: An Open Dataset of Human-Labeled Sound Events. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 30, 829–852.