This is just getting better and better! You did exactly what I was thinking about after getting to know your work through a TGP post. Congratulations on the awesome work!
Good step forward! Now it's complicated for not pro IT programmers to understand how to make VST of your neural models and how to create their own from dry and wet wav. I think there will be leap forward if there will be premaked application - interface with ability to choose wav, train model and create VST
Are you serious, NAM can do captures with authentic knobs? None of the others can do this?? If this is true, it is a revolution. Can you do a video showing you turning those knobs / parameters and how it affects the sound? Why don't I find info about this anywhere else? I'm sure a lot of people would jump on this
Guitar ML Proteus does also "1 knob capture" profiles, but it's limited only to 1 knob AFAIK. Very cool stuff nonetheless, especially for amp profiles (ex : Mesa Boogie Rectifier). It is like the profile version of the Jens Bogren Digital 1 knob (ampsim) plugins.
One question: it's possible to treat other stuff (like a channel switching) as a kind of a "knob" using this approach? If this can be done it could augment a model by a lot, and be even closer to the real thing on a UI perspective.
Yep! Just call off "0" and on "1". The UI only has knobs at the moment so you should turn it all the way in both directions to get what you modeled... Or you could always ask the model what it thinks a "bright switch that's half-on" sounds like! (But no guarantees what it'll sound like! I haven't had time to look into it myself)
I saw this in the TGP thread, however there was so much to wade through and I had very limited time, so I couldn’t read enough about it. My understanding from what I managed to read is that you need to be able to understand coding to get this to work?? Is that still true with this release?? I do not code at all. Thanks