← DatasetsTexte vers audio (musique, sons), Classification et tagging audio, Audio vers audio · mis à jour il y a 1 mois
GH
gnakalehacker/coupe-decale-instrumental-v0-1
⌗ datasetMétadonnées et annotations d’un corpus de coupé-décalé ivoirien : 236 morceaux, 1 838 segments, 15,49 h analysées, avec BPM, énergie, caractéristiques rythmiques, tags audio, captions FR/EN et splits train/validation/test.
Vues
72
Téléchargements
66
Likes
0
Clones
0
Fichiers
10
Taille
5,6 Mo
| # | entier 110 | texte · longueurs 15216 |
|---|---|---|
1 | # Audio rights and redistribution | |
2 | The source corpus was collected from publicly accessible online music sources. | |
3 | Public accessibility does not itself grant redistribution or model-training rights. | |
4 | Before publishing audio files publicly, each recording should have an appropriate rights status, such as: | |
5 | - public domain | |
6 | - compatible Creative Commons license | |
7 | - explicit permission | |
8 | - licensed redistribution/training rights | |
9 | Where rights are unknown or restricted, Mandia should publish only metadata, annotations, derived statistics, fingerprints where appropriate, and dataset-building code rather than redistributing the recording itself. | |
10 | A future release should maintain a per-track `license_status` field. |