Dude That's Wholesome

Yeah, this voice has got a very noticeable static sound in the background. Would sound pretty good if not for that.

Tend to have this issue when using UVR generally, am not an audio engineer so don't really know how to remove it.

A good portion of my models which used UVR also experience this, where the vocal extraction isn't completely perfect.

If you're able to find a solution for me I'd massively appreciate it :) <@358447462782074881>

Which UVR model do you use to separate the vocals from background noise?

I used MDX-NET Inst HQ 1

tried using ensemble mode but it made the extraction process very time consuming for larger datasets

I'd try using MDX23C-InstVoc HQ 2

It's really good

And of course you should manually edit your audio after using uvr to get rid of anything that UVR didn't catch.

And sometimes, the training process gets fucked up in some way and makes your model a little scuffed even if your audio is really good. Sometimes you just gotta retrain it.

but like, at this point I try to avoid using UVR in the first place by just finding audio with no background noise besides the voice. First thing you can do to try and get a great voice.

But of course you can't always find that.

Yeah that static sound is found in the entire dataset

Ah, yeah that's just a UVR model issue.

Nearly all my models are plagued by a static related issue and honestly I do not have the money or time to retrain them all

sadge

I'll just make sure with future models they don't have noise related issues