What's new
By:
Filters
LiteRECORDS

Register a free account today to become a member! Once signed in, you'll be able to participate on this site by adding your own topics and posts, as well as connect with other members through your own private inbox!

  • Guest, before your account can be reviewed you must click the activation link sent to your email account. Please ensure you check your junk folders.
    If you do not see the link after 24 hours please open a support ticket.

2024 best way to separate tracks ?

A

avalon

Executive Producer
hi all
it really feels we are in the future ! not long ago it would have been a dream to think what we can do now with audio (and a lot more .. yes)
what is the best program / procedure to have the best separation ?
what do you all use ?
i thought izotope rx10 music rebalance was the best technology we had ... but im sure there is more
keen to hear what you all use
 
avalon said:
hi all
it really feels we are in the future ! not long ago it would have been a dream to think what we can do now with audio (and a lot more .. yes)
what is the best program / procedure to have the best separation ?
what do you all use ?
i thought izotope rx10 music rebalance was the best technology we had ... but im sure there is more
keen to hear what you all use
Click to expand...

For vocals "MDX23C" and "MelBand Roformer" which can be utilized for free at mvsep.com
For Bass, Drums and Piano "GSEP" also free at gaudiolab.com

Also "Ultimate Vocal Remover"
 
Ultimate Vocal Remover is awesome. They also add new models from time to time. We live in great times. :)
 
I also use mvsep.com

You need to make a free login to get the .flac option, and also with a login you get a bit ahead in the "que".

This is my favorit settings on mvsep.com:

For 6 channel multi tack extraction:
multi.jpg

For vocal extraction:
karaoke.jpg
 
Last edited:
One of the large problems with standalone software for stem isolation is that they use whatever AI model was out at the time of development. It's best to use sites like:

  • lalala.ai
  • mvsep.com
  • x-minus.pro/ai

Or use the original UVR vocal remover python/colab program but use the latest models.
Some models include:
  • DEMUCS
  • UVR
  • MDX
  • RoFormer
and each has many versions which work better on different input audios.

My favorite site so far is x-minus.pro/ai since they keep on top of the latest models, but it requires payment for best results.
 
I recently came across a free offline AI tool for generating stems from music files (or by searching YouTube from within the UI): StemRoller
It uses the Demucs AI algorithm, developed by Facebook.

I'm mainly working with metal music, so you might get much better results if your music is less cacophonous! It's great for getting drums and clean vocals, but not much help with bass at all, and guitars that use distortion are kinda iffy unless they're conveniently panned to the extreme left and right. But the drum and clean vocal output is really solid, even for low-quality metal productions from the '80s.

Overall it's a great wee program. And being free and offline, you don't need to sign up for AI credits or limit your output to 3 minutes etc. I've compared the output quality to a few online stem generators, and the results are very close to, and often better than, the results those websites give. As with all things AI, your mileage will absolutely vary!
 
You must log in or register to reply here.
Back
Top