EXAMPLE Neural Seed: Running Neural Networks on Daisy Seed for Amp/Pedal Emulation

[keyth72]

If anyone is interested in using A.I. tech on the Daisy Seed, here is my example project using a neural net to emulate real amps and pedals. The code is all open source on Github:

GitHub - GuitarML/NeuralSeed: Neural networks for guitar amp/pedal emulation on Daisy Seed

Neural networks for guitar amp/pedal emulation on Daisy Seed - GitHub - GuitarML/NeuralSeed: Neural networks for guitar amp/pedal emulation on Daisy Seed

github.com

I posted my Neural Seed project in the Build Reports section earlier but missed this Electro-Smith category, here is the original post:

DEMO - Neural Seed: A.I. Amp/Pedal Capture on the Terrarium Pedal

I just finished a project using the Terrarium pedal (powered by the Daisy Seed digital board), which uses neural networks to emulate real amps and pedals. Similar in concept to Kemper/ToneX/NeuralDSP but scaled down to an inexpensive DIY platform (total cost was around $100). The code is open...

forum.pedalpcb.com

I figure this thread can be for discussion of the technical details, if anyone is interested. Neural Seed uses the open source RTNeural engine to run pre-trained amp and pedal models. It can run models of clean or low gain amps, and overdrive/distortion pedals. Model training is accomplished using Pytorch. Here is the main c++ code, which runs the real-time neural models, and associated audio controls for the pedal:

NeuralSeed/NeuralSeed/neuralseed.cpp at main · GuitarML/NeuralSeed

Neural networks for guitar amp/pedal emulation on Daisy Seed - GuitarML/NeuralSeed

github.com

 

[KR Sound]

Definitely checking this out.

 

[giovanni]

I just read your article on Towards Data Science. Very fascinating stuff! I have been working with ML professionally for the last 8 years and I never imagined my hobby would intersect with work stuff Very inspiring work man!