The upgraded version 3 of Facebook's Demucs unmixing tool is one of the best open-source solution at the moment: https://github.com/adefossez/mdx21_demucs/tree/track_b.
The "MDX B (UVR)" model is available here: https://github.com/Anjok07/ultimatevocalremovergui/releases/tag/MDX-Net

 

moises.ai/ this is one I know. Haven't tried any others here but the results really surprised me. You can do 5 songs a month and Audio will be cut of at 11 kHz on the free plan. Download the audio labeled high res from the email they send you the quality iy much better than the one from the website.

 

I been playing with Spleeter this afternoon and the result is almost the same as what I get from izotope. VERY GOOD !
But the best result I could get (tried 3 different songs) was 3/3 from Steinberg Spectralayers.

Precious tips

 

Oh, and one more thing, Zynaptiq's Unchirp can make things sound a bit more natural after all that processing, worth a shot.

 

looks promising, but unfortunately you need to be a software engineer to figure out how to install and use this. lol

 

Exactly! I gave up after few minutes spent totally lost on that page!

 

may i ask which version of spectralayers you are using? i read that there were problems with vr's version 8.

 

Steinberg.SpectraLayers.Pro.v8.0 VR downloaded from the sister site. Didn't encounter any problems using it.

 

Unfortunately all links are dead

 

It's still out there somewhere, PM'd.

 

UVR (Ultimate Vocal Remover) was already mentioned and even though it sounds like a joke, I consider it to be the best option for overall removal/separation of vocals.
Vocal tracks that recent versions of RX, SpectraLayers or Spleeter have hard time separating (namely Au5 - Snowblind and Tristam & Braken - Frame of Mind as good examples for notoriously hard to extract vocals) are coming out way cleaner using the models available for UVR.
It takes a bit of time to set up and get everything running, but after that it's smooth and most-importantly free sailing.

 

Don't know it but most def not a joke for me.
Uses another AI with added GUI, in the same line as the Spleeter one commented before.

 

Izoptopes Rx Adavnced Editor and a lot of Time does ist.
There is no magic plugin that 'just does it'.
You need to paint it out in the fft ananylzed graphs.
(edit: that magically does it in any usable way, that is..)

 

check out this one. according to users, it is better than lalal.ai and the free plan gives you a lot more. but you can sign up for premium for as little as $1 per month: https://x-minus.pro/ai

 

Youtube-dl and demucs makes for a great karaoke party!

 

I have always thought this plugin, Peel from zplane, was cool. It may be better or worse than the other plugins/programs mentioned above, I don't know/haven't tried hem all except RX which is great

https://products.zplane.de/products/peel/

One of the features is being able to route difference frequency ranges/instruments to separate tracks for specific processing.

 

It did not sound like a joke at all. But it looks very complicated to setup! I gave up but you tempted me to try again

 

I use https://moises.ai/ all the time with great results .Just use a fake email so you can bypass the track limit

 

Downloaded! Will give it a try tomorrow

 

If you're looking for something you can use offline, AconDigital's Acoustica wave editor has an unmixer function that I think uses, or is at least based on, the Spleeter AI.
You have to download it separately after the install (it's part of the installer) but I've found you can install version from the sister site (or the trial version from their site), download the unmix add on, then apply the R2R keygen and disconnect from the internet (or block it in your firewall).
Latest version on the sister site is 7.2.8 but I've done it with 7.3.0 using that keygen.