I just want to hang this idea out here, one idea reason is for me to collect my thoughts so that it’s not in shambles, so I can go back to check out what I want. Another reason would be for someone other then me to take the idea and run with it.
OK, so this is what I want:
* An AU/VST plugin that can load a settings file (ie neural network) that emulates an effect (be it an old vintage distortion or something other out there)
This is how I think it could be done:
* Make a neural network trainer where you input the original audio file, what kind of parameters to run on it, and the expected result. Then then train the neural network with new output and change the parameters to that that would output the new sound-file.
A good thing about being a kickstarter junkie is that open sometimes buy’s really really useful stuff like this: http://www.shiffman.net/teaching/nature/nn/ but in .pdf format, and with even more info. So maybe, maybe I will be able to make this happen even though I previously thought I lacked the math skills needed.
Now if you’lll excuse me I’m gonna do a whole lot of reading.