Hi,

I am looking for a free AI splitter similar to FADR, or possibly some other software
What is your experience with this? what do you think is the best on the market?

thanks!

 

UVR -> The Ultimate Vocal Remover Application
Hence the name, it can also do 4 and 6-stem separation
It has really active and helpful community in discord

 

A very good explanation of the topic. Some good examples of how different versions free and paid compare

As a side note And nothing to do with what you're looking for, but I found this interesting.
It's a piad thing but still cool to look at.

 

Last edited: Mar 23, 2024

Unless you want to waste 1h30 of your time, 100% useless video:
- It was published last week but all these are outdated & outperformed splitting algos
- Spleeter is outdated, outperformed and is consireded as crap since 5 years.
- The "tester" doesn't even realise that he's testing the same model multiple times as Spectralayers, Fadr and DemucsGUI are using same Demucs models.
- "Ai-lyd.no" looks like scam.


Join the Audio Separation discord community Legotron has already mentionned, you'll learn more than with clickbait bullshit videos.

 

Last edited by a moderator: Mar 23, 2024

A new UVR beta has been released in Audio Separation discord with new models based on ByteDance's algo which was reproduced from scratched an open-sourced recently.

Highest SDR score we've ever seen for vocals/instrumental stem.

These models will be implemented in few days in upcoming update of my colab fork, with some other optimisations.

They are also already available on mvsep.com & x-minus.pro (UVR-Online)

 

Wait... Which Demucs model separates 14 stems from Music like Fadr? Or if that's another model, how do I use it? Thank you.

 

I want to love this and be an active user/participant, but I'm really having trouble getting started. Do you have a link for the Discord community? Any advice? There are so many models to download . . . me confused.

Is this it?

https://discord.com/channels/708579735583588363/708595418400817162

 

Last edited: Nov 30, 2024

@tommyzai Yes, that´s the one
Just ask guidance there, there are also documents for different usage of models

 

I use Fadr, its great, however there are possibly better options as pointed above. Fadr is also on steep sale for annual right now. 35$. It also comes with unlimited cloud storage, a splitting plugin, and a "synthGpt" plugin, which is text to synth plugin. type what sound you want like "dreamy pad", or "big 808" etc, and it generates the sound for you and you can play it like a synth. Its decent. Not replacing any synth for me yet, but its fun to play with or generate stuff for resampling. Will likely get better with time, as they seem to be constantly updating.