Integrating ML Stuff?

[iamlamprey]

How hard would it be to integrate/recreate something like this in HISE? It's basically an Additive Synth, Subtractive Noise and Convolution, with the parameters being called from the trained models

https://arxiv.org/pdf/2103.07220.pdf
https://github.com/SMC704/juce-ddsp

I've seen some other ML vst's out there using JUCE and importing the trained model data as .JSON, I think there's some cool applications for Machine Learning and audio