I really preferred UVR to the other clunky one which is commercial software and needs to be k'd, cant remember the name.
But the development hasnt moved since September 2023, there were some MACOS commits since then i think until October 2023.
I was looking forward to hearing the results get better and better but it seems dead in the water?

Anyone know whats going on?.. Anjok07 seems to be in charge of the project, maybe someone else made something better

 

I have a very old video card that it is probably not fair to even call it a gpu. I use Hit N' Mix RipX on the rare occasion I need to stem separate anything and it works great for my purposes. It's not fast or at the top of a leaderboard, but it works very nicely compared with what I originally expected of it. It is most likely the program you mentioned.

 

ensamble mode is really good in UVR. There are a lot of 3rd party algos you can run in there MDX23 is one of them

 

I just entered a "BATTLE OF THE AI BANDS" and needed proper separation for AI-generated tracks. Now Im just starting to really dig into all the different 3rd party algos with UVR5 and chaining them in ensemble mode. I was just comparing GUIs like the rest of the clueless before I knew what was really going on under the hood with the AI models

 

going to check out spectralayers, i read someone saying it was better quality output than UVR

 

Longtime lurker, first time posting. FWIW, UVR is still being maintained and there is a active community on discord.

 

There is a difference between SL and UVR, and I mean from SL 11 because v10 was not even close to UVR.
Now the difference is huge, UVR still doing better work without tweaking and long render time waiting.
In most parts as I need just two of stems vocal and instrumental, UVR is a workhorse, SL is more complex and render times is awful and still UVR is doing far better job than SL 11.
But for sure SL 11 have huge improvements from v10, now UVR and SL are quite comparable, but UVR was great discovery for me since last year and still my way to go for vocal removing.
Now regarding those modes and algos I don't know nothing, I never found a proper explanation how that works so I use the default settings, maybe there is better tweaking solution and fine tuning, but I don't know how to use that.
Just My thoughts.

 

It is my understanding that SL 11 uses a fined tuned demucs model, the same type that UVR has, only Steinberg's is tailored to SL. If interested, mvsep.com is the web version of UVR, plus it has the BS/Mel Roformer models. It's free, however if you want .WAV or .FLAC outputs, you'll need to register. Ensembles cost $ and there is a queue that sometimes is rather long.

 

Interesting. There are a ton of reports to the Dev that (even though older versions worked perfectly fine) updates to the program cause it to refuse to open on tons of different Mac computers with various OS versions.

Not saying the dev is obligated to fix the issues but I can personally say that on about 4 Macs I've tried the program on, only 1 actually worked and it stopped working when I updated to a new version of UVR.

Is there a way to access the older version of UVR? I haven't found an archive of the older versions and to my knowledge they have been removed from the GitHub page.

 

You're right, I've seen posts about the Mac issues, however, I don't use a Mac so I pay no attention to anything regarding it. As for the older versions, how much older are you interested in? And I suppose it is for Mac?

EDIT: Here is a link to the Google doc that is frequently updated with all things UVR and its counterparts and then some. It's REALLY long but very informative.
https://docs.google.com/document/d/...bAxEh_OBv94ZdRG5c/edit#heading=h.roiuj54hzww3

 

Last edited: Sep 9, 2024

I used UVR and hit n Mix both but to my surprise the results were one of the same. May be because i was trying really difficult tracks to stems. But i found that, if you don't care about instruments and only vocal stem. Moises application is really a good one not only it gives good results in complex tracks but also gives bpm and key. Which is otherwise difficult to accurately detect in complex song structures

 

Would love the discord link pls <3

 

No problem!
https://discord.gg/ZPtAU5R6rP

 

The last i used virtual DJ was in 2007. When it used to come under Atomix label. Not sure if it still does. But yeah it was my first encounter towards music mixing and other stuff. Those were good old days.

 

because it's flawless as it is, and models can be downloaded within its GUI (in settings page)

 

Can you name some of the 3rd party algos please? Looking for more