Oh OK. I didn't know if this was updateable or not.
Printable View
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.
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..
There already is a Colab for it (the link is from the github page)
https://colab.research.google.com/dr...GH?usp=sharing