Not all music is well represented with notes, nor are audio datasets with high-quality note representations readily available. But I guess if you work hard enough you can get close: https://www.youtube.com/watch?v=o5aeuhad3OM My example still sounds like the chiptune simulation that it is, however.