Hi! Here's my question : is Synplant2's Phenotype LLM trainable by me for patches for my audio modeling synths like Chromophone, Atom, and Anukari , which would take in signals probably in OSC that are more abstract versions of what I'll really be sending in an encoded way via MPE? I'm pretty uninterested in the standard synth wave forms, envelopes, LFOs and most filters.

Or a less complicated question: I have a Mirage running Soundprocess (yes, 80s -90s tech) which can have its nasty bunch of parameters changed in real time, so training on a set that understands that and then can use that training to do realtime timbral changes via CCs on this device would be great. basically, a nice working LLM wrapper around training and generation taking descriptive parameters and abstracted runtime parameters that result in CC changes and Sysexes and an occasional wavetable without the misery of LLM dependency patching , wrapped up in a plugin for real time and an offline training app is what I'm looking for.

I mean, I can write this, but I'd rather someone who's already partly there take it on.
Not much to ask, really, right?