So that raises an interesting question, since you did it to your own track - how do the extracted parts compare to the original tracks? Assuming you still have the project it should be easy to A/B them and even run some analysis. For whatever reason this didnât occur to me when I was testing it, probably because I was just playing around and not really deep diving, but it might be illustrative to compare.
I guess on balance it doesnât really matter. 99% of the time youâll be throwing someone elseâs work in there and it sounds like what comes out would be useful, so whatever. Iâm just curious how well it did when compared to the original.
Thatâs a great idea. I added the original stems to the playlist and made them downloadable. I donât have time right now at this very moment but ideally Iâd love to get this up into audacity to really check out the spectrogram for comparisons. I wonder if ChatGPT is able to compare as well?
Just a quick listen I can certainly hear artifacts and under upper mid compression in the vox A/B, but honestly itâs still quite precise, to my ears on first listen.
edit: I realized just now that I stemmed the Bass and Perc together on the Originals render because the âbassâ is just a tuned sub kick, so just combine the Ableton Drums/Bass stems.
Holdup, wasnât this already out for a few months, or am I MacDonalds special sauce? I recall watching a video about it and being kinda hyped for no reason - maybe that was a beta build or somethingâŠ