Any news with spleeter or its models? It seems that the project has been abandoned.

[titet11]

I think the original programmer of spleeter abandoned the project, since there has been no news about it. Also, I found that the following site is better for voice isolation than the program itself:

https://vocalremover.org/es/

[mdolnik]

Also Audacity now has a plugin for AI audio separation:
https://www.audacityteam.org/blog/openvino-ai-effects/#ai-tools-for-music

[titet11]

@mdolnik I already knew it, but I think this application had more future to improve than openvino.

[boy1dr]

The last commit on the Deezer Spleeter repo is over a year old now (Jul 13, 2023). it's safe to say Spleeter is as good as it's going to get.
This form of separation was never going to achieve high fidelity anyway, not without some kind of synthesis to fill in the gaps or artifacts. It is more than likely this will be revisited again in the future when ai generative audio can replace existing sounds.
Deezer's reason for creating this project was to identify Music IP infringements, it is not likely Deezer require it to sound good. But i thank for their awesome contribution :)
It would have been nice to see an official Windows GUI version though.

[titet11]

@boy1dr I know this project had poor fidelity with karaoke, but its interface was pretty good.

[boy1dr]

@titet11 there's an unreleased version of SpleeterGUI, i could open it up with a notice that it is an unfinished project.
It was working but language translations hadn't been implemented.
Nicer interface, non-blocking runs...just keep dumping files in to it.
It was to be version 3 which is why there where so many changes to version numbers towards the end there.
of interest or too late never mind?

[titet11]

@boy1dr Well, then I would like to ask you the following question:

So far, could there be projects similar to spleeter that could be used in the program?

[titet11]

@boy1dr

Do you think that the open vino project could implement it here, it has better voice isolations.