After reading the readme I'm not sure how this works. Will the "prompt" be the "raw audio" of the song? And the model returns lyrics, chords, beat, melody and the tabs?

I'm looking for something that I can sing to (a self composed song without music) and it would recommend guitar cords/tabs + melody to accompany myself. Any ideas?