Stable Audio Open was trained on Creative Commons-licensed audio data from Freesound and the Free Music Archive.
