I never would have thought it possible, but here we are. This is a track a friend of mine recorded on a Yamaha MT1X 4 track cassette machine some 38 years ago or so… he’d always wanted to remix it, but the Yamaha is long gone. I decided to take it on as a challenge when I was able to very successfully separate stereo files into separate stems using voice.ai
It took a bit of doing and a few editing tricks but here is what I ended up with:
…and here’s the original 1986 mix for comparison :
Well done! So much cleaner. Pretty crazy how powerful the tools are nowadays.
Also, I hear some primus influence in that song! Especially lyrically. Good stuff 
Thanks - It was a LOT of work, but still, just a year ago it would have been absolutely impossible.
Definitely no Primus influence, though. These guys are true originals. Tim (the lead vocalist and bassist) didn’t need any outside influence to be this weird - he and Steve (guitarist) were born that way! Haha! Officially they’re known as “Loose Change”, but early name choices included “Fungus Wardrobe”.
Besides the band being from the other side of the world (Australia), these guys were going in the mid 80s, and Primus didn’t even release their first album until 1989… & we certainly didn’t hear of Primus here in Oz until the 90s.
… Funny, the similarity never occurred to me, but now that you mention it, I totally get why you might think that!
That’s very impressive @ColdRoomStudio , and the band is cool. Thanks for sharing this. Just curious, you say it was a lot of work, after the AI separates the tracks isn’t it just a regular mix session?
Thanks! Yes, I’m amazed that it turned out as well as it did!
This wasn’t quite as simple "split the stems and rebalance them, but then again, that wasn’t my goal - I really wanted it to sound good to a modern listener. With that in mind, I did considerable extra work to get the result I did.
I stitched together a second guitar track from the existing guitar track to create a double track for the guitars. I also duplicated the drum track, split the kick, snare and tom hits using tab-to-transient, and triggered close mic drum samples from the split up tracks to give the whole thing a fuller sound. Apart from that it was pretty straight-forward mixing stuff.
My suggestion is to try voice.ai out for yourself. Go to the “online tools” section of their website, grab an .mp3 of one of your mixes and load it up in the “Stem Splitter” page here: Free Online Ai Stem Splitter - Isolate instruments from a song - Voice.ai - Check out the results you get.
The tools are free to use with .mp3, but to use .wavs and download them you need to pay a monthly subscription fee - I Think about $5 IIRC.