Skip to content

Dataset · benjamin-paine

Cleaned FreeSound audio dataset from LAION (630k+)

benjamin-paine/freesound-laion-640k

Re-uploaded version of the LAION-Audio-630k audio dataset with text descriptions of sounds and verified licenses.

What's inside

Contains between 100 thousand and 1 million audio recordings with basic metadata and authorship information, taken from the FreeSound.org collection by the LAION project. The repository authors kept only the audio itself and basic columns, added license data, removed download links to save space, and deleted about 1,000 samples with unclear license status.

What you can do with it

  • Training sound classification and acoustic event models
  • Training text-to-audio and audio-to-audio sound effect generation models
  • Training audio encoders for multimodal systems
Tasks
Audio → audio, Audio classification
Size
100K–1M rows
Downloads this month
33.8K
Likes
14

How to load it

from datasets import load_dataset

ds = load_dataset("benjamin-paine/freesound-laion-640k")