The actual effect transforming the audio is a neural network. The beauty is that you can record someone playing guitar through an ultra-rare amp/filter/whatnot (both clean signal and output) and then train the neural network to replicate that.