GuitarML / PrincePedal

Prince of Tone style guitar plugin made with neural networks

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

The Prince

CI License: GPL v3 Downloads

The Prince is a plugin of my homebuilt Prince of Tone style pedal, cloned using neural networks. The graphics were created from actual photos of my pedal using a "stop motion" technique (not perfect but it works). The plugin features three GuitarML neural network models conditioned on the Gain and Tone knobs, one each for Overdrive, Boost, and Distortion modes. The Prince should be used with an impulse response plugin (such as Pulse) to emulate playing the pedal through an amplifier. The original Prince of Tone pedal is essentially 1/2 of the highly sought after King of Tone by AnalogMan. Use two instances of the Prince for a King of Tone - like experience!

app

This plugin is intended as an example of how a pedal/amp builder (hobbyist or professional) can faithfully recreate their work in the digital world using neural networks.

Installing the plugin

  1. Download the appropriate plugin installer here (Windows, Mac, Linux)
  2. Run the installer and follow the instructions. May need to reboot to allow your DAW to recognize the new plugin.

Info

The Automated-GuitarAmpModelling project was used to train the .json models.
GuitarML maintains a fork with a few extra helpful features, including a Colab training script and wav file processing for conditioned parameters.

The plugin uses RTNeural, which is a highly optimized neural net inference engine intended for audio applications.

For the training data, five steps for the gain and tone knobs were recorded (0.0, 0.25, 0.50, 0.75, 1.0), for three modes for a total of 75 wav samples at 2 minutes 20 seconds each. The Proteus capture utility was used for training, with a modified config for two knobs (Gain and Tone). The training data was normalized, so the volume on the plugin is more consistent between modes than on the actual pedal.

Note: There is no stompswitch bypass button, which is different from other pedal style GuitarML plugins, it is simply part of the background image. Use the DAW to control bypassing the plugin.

Pedal Build and Differences

app

I built this pedal in a 1590B enclosure with custom "paint pour" technique by a talented local artist. The circuit was soldered on a 2 sided through hole PCB, and is powered by a typical 9v power supply (no battery). The circuit design is a unique take on the Blues Breaker style circuit with two op amp gain stages and silicon clipping diodes. The three modes are determined by the symmetric silicon clipping diodes. Overdrive uses 4 "soft clipping" diodes within the second op amp feedback stage for a smoother sound. Distortion uses two "hard clipping" diodes after the second op amp, clipped to reference voltage. Boost removes the clipping diodes, for a more open sound with slight distortion from the op amp.

There are several differences between the Prince of Tone pedal by AnalogMan and my homebuilt clone, mainly due to available parts:

  • My homebuilt pedal is technically 1/2 of the King of Tone, so no internal dip switches for Low-Mid and Turbo
  • 20k tone pot instead of 25k, so slightly less range on the bass side of the tone knob
  • 22k resistor instead of 27k on the first op amp filter stage
  • TL082 IC instead of JRC4580
  • The internal Treble trim pot was set to 50% for all my training recordings
  • No knob or switch labeling (didn't want to obscure the beautiful paint pour art)
  • A white LED brighter than the sun

Build Instructions

Build with Cmake

# Clone the repository
$ git clone https://github.com/GuitarML/PrincePedal.git
$ cd PrincePedal

# initialize and set up submodules
$ git submodule update --init --recursive

# build with CMake
$ cmake -Bbuild
$ cmake --build build --config Release

The binaries will be located in PrincePedal/build/Prince_artefacts/

About

Prince of Tone style guitar plugin made with neural networks

License:GNU General Public License v3.0


Languages

Language:C++ 55.8%Language:Shell 21.7%Language:CMake 14.6%Language:Inno Setup 7.9%