Audio vocal/instrument separator

[k95555]

I am trying to separate the vocals and instruments using UVR-MDX-Voc_FT.onnx. The issue I am facing is that this model does not separate the instruments effectively. I can still hear the vocals playing very quietly. It does not remove the vocals completely. I tried the other two models in the dropdown menu, but they are even worse or not intended for this purpose. I had an older version of RVC, which included the hp2.pth and hp3.pth models, and they were extremely good at separating vocals. I tried downloading other .onnx models and placing them in the vocal separator folder, but they do not appear in the dropdown menu even after restarting. It looks like these three models are hardcoded with Ultimate RVC. Has anyone else encountered this issue with this separator model and found a solution on how to add new models to the directory? I would greatly appreciate any input.

[JackismyShephard]

Hi @k95555

the audio separation models are hardcoded now. I will consider adding more in the future. Maybe I will consider these hp models but I think in general they might be outdated?

[k95555]

Hi @k95555

the audio separation models are hardcoded now. I will consider adding more in the future. Maybe I will consider these hp models but I think in general they might be outdated?

Thank you so much for the quick reply and thank you for this ultimate tool, It's great and the step-by-step production is clean and well thought out. I am not sure if HP models are outdated. At the moment, I am still using the old RVC just for separating vocals, then I process them through Ultimate RVC. To be honest, HP2 is quite good at separating vocals and instruments. It does a very clean job. There are a few more models in old rvc, such as Deverb and Deecho, that are quite good too.

[JackismyShephard]

@k95555 Thank you for the positive feedback. Its great to know that people are actually using the multi-step functionality as that is something I worked hard on implementing.

What do you mean by the old RVC? Can you supply a link? if these separation models are superior its definitely something we should consider implementing. Right now I am a bit busy with other projects so don't have that much time for Ultimate RVC. I will make this a priority though when I have more time.

[k95555]

@k95555 Thank you for the positive feedback. Its great to know that people are actually using the multi-step functionality as that is something I worked hard on implementing.

What do you mean by the old RVC? Can you supply a link? if these separation models are superior its definitely something we should consider implementing. Right now I am a bit busy with other projects so don't have that much time for Ultimate RVC. I will make this a priority though when I have more time.

old rvc ,i mean the one was using before. https://huggingface.co/lj1995/VoiceConversionWebUI/tree/main ,it has good vocal separation models ,if you can add these would be great or maybe change it so we can download and add models into the directory and yea that multi step is great and very intuitive.no doubt its must been a lot of work.

[k95555]

@k95555 Thank you for the positive feedback. Its great to know that people are actually using the multi-step functionality as that is something I worked hard on implementing.

What do you mean by the old RVC? Can you supply a link? if these separation models are superior its definitely something we should consider implementing. Right now I am a bit busy with other projects so don't have that much time for Ultimate RVC. I will make this a priority though when I have more time.

i understand you must be very busy,but if you get time.update and add new models to it,please leave comment here ,i will highly appreciate .thank a lot