Is there a headless mode, for synthesizing sounds and performing FX, from an API? Or is this like 99.9% of music software that the authors assume only a human will be the user?
I've been doing research on music AI, inverse synthesis, and the like, and shockingly few open-source software packages were usable for creating training sets.
There's a scripting interface[1] which might help you do some of what you want, but I'm not sure if patches can be run headless.
If you haven't seen them already, SuperCollider, Chuck, PD, Faust and/or the Web Audio API might be better suited to generating training sets in an offline fashion.
I've been doing research on music AI, inverse synthesis, and the like, and shockingly few open-source software packages were usable for creating training sets.