Can any one suggest me how to create custom dataset for sound source seperation task from a given mixed signal?

[Istiak665]

I want to create a deep-learning model that can perform the mixed signal separation. For this case, I need suggestions on which model can be used and how to create the custom data set for that specific model. Thanks in advanced everyone.

I have already tried to create and use some developed models like u-wave-net, open-unmix, and u-net. But I can understand how to create custom datasets as well as model creation.