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.

Demucs - Stems Extraction Tool

joaoasq said:
Comparing demucs with Ultimate vocal remover...

UVR is way better, for making instrumentals

Demucs:
I noticed a lot of remaining vocals in the "drums" stem.
And it seems the background vocals are left in the synth stem.
Click to expand...

You should hear how good and clear Alice in Chains - Nutshell (Live) MTV Unplugged turned out as a pella and mental. Clean as hell.
 
NewAgeRipper said:
Am I missing something with this? I added Nutshell.flac" --shifts=5 and it only ran the process once, but slightly slower.
Click to expand...

That's it, it doesn't do passes like UVR. I think this shifts option mainly reduces the hissing that all of the stems get, maybe if you listen to them on headphones you'll know what I'm talking about.
 
djtayz said:
That's it, it doesn't do passes like UVR. I think this shifts option mainly reduces the hissing that all of the stems get, maybe if you listen to them on headphones you'll know what I'm talking about.
Click to expand...

I did. I figured it was doing something like that. It wasn't that I was ditzing the process, I just wanted to be sure I was doing it right.
 
Angelman60 said:
Thanks guys, and it's true that step 2 is a bit long like 12 minutes here. Is there a way to process 2 or 3 tracks in a row without repeating the first steps ?
Click to expand...

even when u eventually turn it off and need to do the steps all over again.........
installing the ffmpeg....hit it...let it run for a minute or so and hit it again.....
should install no problems
 
rkeane said:
even when u eventually turn it off and need to do the steps all over again.........
installing the ffmpeg....hit it...let it run for a minute or so and hit it again.....
should install no problems
Click to expand...

If you accidentally turn it off and reopen it in the next 3-5 minutes, you should be fine without redoing all the steps. Just see what works best for you.
There's a "glitch" to keep it active all the time but I'm not sure if it's worth posting it.
 
rkeane said:
even when u eventually turn it off and need to do the steps all over again.........
installing the ffmpeg....hit it...let it run for a minute or so and hit it again.....
should install no problems
Click to expand...

Yep. That works for me. Start ffmpeg. 1 min later, interrupt it. Keep doing rest of steps. Thanks for the tip.
 
Is there any trainable models being developed to strictly make high quality vocal extractions from a song? Seems the vocal isolating technology still has a long ways to go to catch up to vocal removing.
 
INGSOC said:
Is there any trainable models being developed to strictly make high quality vocal extractions from a song? Seems the vocal isolating technology still has a long ways to go to catch up to vocal removing.
Click to expand...

There is a guide on how to train a model on the GitHub page. They have also listed the sources of music they used to train the default models
 
djtayz said:
What do you mean? There aren't new models for demucs to make an update
Click to expand...

I found out about the updated Demucs 3 with Model A & B option and the only way I can use it is by using some website in Russia because they offer it for free. I really don't like using a website hosted there because of the current situation which I don't think really needs explaining, don't get me wrong I love the Russian people and the ones not falling for the propaganda but I have issue with using the site. That all being said I would gladly pay you if you could make a new colab for the demucs 3 A & B model. I have ran a couple of songs through it and cant believe how quickly this A.I. tech is coming along. I know you're very busy but please consider this... All the best..
 
chilinvilin said:
I found out about the updated Demucs 3 with Model A & B option and the only way I can use it is by using some website in Russia because they offer it for free. I really don't like using a website hosted there because of the current situation which I don't think really needs explaining, don't get me wrong I love the Russian people and the ones not falling for the propaganda but I have issue with using the site. That all being said I would gladly pay you if you could make a new colab for the demucs 3 A & B model. I have ran a couple of songs through it and cant believe how quickly this A.I. tech is coming along. I know you're very busy but please consider this... All the best..
Click to expand...

There already is a Colab for it (the link is from the github page)

https://colab.research.google.com/drive/1dC9nVxk3V_VPjUADsnFu8EiT-xnU1tGH?usp=sharing
 
You must log in or register to reply here.