If you want to do manual seperate ;

[Enes-s]

If you want to do manual seperate ;
It can be integrated into the sound obtained at the model output or the input sound. There is a seperate code inside but it can be used if you want to do it more manually.
I think it is more effective than the one in colab

from audio_separator import Separator

Initialize the Separator with the audio file and model name

seperate_target_audio = "/content/1.wav" #@param {type:"string"}
separator = Separator(seperate_target_audio, model_name='UVR_MDXNET_KARA_2')

Perform the separation

primary_stem_path, secondary_stem_path = separator.separate()

print(f'Primary stem saved at {primary_stem_path}')
print(f'Secondary stem saved at {secondary_stem_path}')