Still using Spleeter and even the sample on there page sucks. Anyone that can use UVR GUI by Anjok or the UVR Google Colab and the demucs colab by Djtayz is far better off since UVR doesn't leave dropped volumes where vocals were removed.
Still using Spleeter and even the sample on there page sucks. Anyone that can use UVR GUI by Anjok or the UVR Google Colab and the demucs colab by Djtayz is far better off since UVR doesn't leave dropped volumes where vocals were removed.
All I can tell you is follow the video for how to set up ffmpeg. That's the biggest part. The rest is running a couple of command lines to get the rest of what UVR needs. No idea about the Colab models. They are still testing many things to determine what will be best and most user friendly.