MEMLNaut Firmware: Multi FX Processor with Reinforement Learning
Overview
This firmware offers exploration of multi fx, using the CARL approach to ML.
Multi Effect Algorithms
The processor has the following effects chain:
Pitch shifter -> Stereo Delay -> Dry/Wet Mix -> Outputs
Output from the neural network controls:
- The pitch shift frequency (from -12 to +12 semitones)
- The delay time, feedback and dry/wet mix for delays on each channel
- The overall dry/wet balance
Inputs
The processor takes a mono input on the right channel.
Outputs
Stereo sound is output to the line outs and headphone sockets.
Reinforcement Learning
When you indicate a reward, you are rewarding (negatively or positively) how a particular joystick position is mapped to the sound it’s making. As you indicate preferences for these settings, the DDPG algorithm will start to tune how the actor reacts to the joystick.
Files
Code
https://github.com/MusicallyEmbodiedML/memlnaut-multifx-carl