original_audio: the chunk cut from the original full recording.
demucs_audio: the same interval cut from the saved full-recording Demucs
vocals MP3.
The rows are the chunks selected by the production music detector. Use the two
players to audit whether Demucs improves speech quality and removes background
music. No chunk-level source separation was performed. Audio is mono 48 kHz,
96 kbps MP3.… See the full description on the dataset page:
https://huggingface.co/datasets/Kppwdfgu1/sbpn-music-demucs-side-by-side-20260722.
