Baike.dev
All toolsOnline toolsAI codingTrendingOpen sourceNewsSubmit
Log in
< Back to tools
K

KlonCentaur

> 编程语言
Open source

Digital emulation of the Klon Centaur guitar pedal using RNNs, Wave Digital Filters, and more

380 stars0 likes0 views
WebsiteGitHub

About

Digital emulation of the Klon Centaur guitar pedal using RNNs, Wave Digital Filters, and more

Klon Centaur Model

This repository contains a digital model of the Klon Centaur guitar pedal. The model is constructed using a variety of circuit modelling techniques, including nodal analysis, wave digital filters, and recurrent neural networks. The model is implemented as an audio plugin (Standalone/VST/AU/LV2) for desktop and iOS, and as a guitar pedal-style effect embedded on a Teensy microcontroller. This work began as part of a class project for EE 292D at Stanford University.

Quick Links:

  • Latest Release
  • ChowCentaur for iOS
  • Nightly Builds
  • Presentation at the 2020 Audio Developer Conference
  • Technical paper on the ArXiv
  • Video Demo

Audio Plugin

The latest release can be downloaded from our website. ChowCentaur for iOS can be downloaded from the app store. The latest builds (potentially unstable) can be found on our Nightly Builds page. Linux users can find builds available on the Open Build Service, courtesy of Konstantin Voinov.

Building from Source

To build the audio plugin, you must have CMake installed (version 3.15 or greater). Then use the following steps:

bash
# clone repository
$ git clone https://github.com/jatinchowdhury18/KlonCentaur.git
$ cd KlonCentaur
$ git submodule update --init --recursive

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

ChowCentaur also has a headless mode that contains a performance benchmarking app. You can run the benchmarks yourself as follows:

bash
# build with CMake
$ cmake -Bbuild -DBUILD_CENTAUR_HEADLESS=ON
$ cmake --build build/ --config Release --target Centaur_Headless

# run benchmarks
$ ./build/ChowCentaurHeadless bench

For more information, run ./build/ChowCentaurHeadless bench --help.

ChowCentaur uses the RTNeural neural network inferencing engine for running computing the output of a recurrent neural network in real-time. RTNeural has three available computational backends: Eigen, xsimd, and STL. By default, ChowCentaur uses the xsimd backend, but that can be changed using a different CMake configuration command, for example: cmake -Bbuild -DRTNEURAL_EIGEN=ON.

Teensy Pedal

Check out the video demo on YouTube! For more information on the Teensy pedal-style implementation, see the TeensyCentaur/ subfolder.

Circuit Modelling

The circuit model is constructed using nodal analysis and wave digital filters. For more information see:

  • Julius Smith, Physical Audio Signal Processing
  • Kurt Werner, Virtual Analog Modelling of Audio Circuitry Using Wave Digital Filters

The wave digital filters are implemented using a WDF library, available here.

Neural Network Modelling

In the neural network version of the emulation, a recurrent neural network is used to emulate the gain stage circuit of the original pedal. The RNN architecture used is derived from the one presented by Wright et. al. in their 2019 DAFx paper "Real-Time Black-Box Modelling with Recurrent Neural Networks". Training data consists of ~4 minutes of Direct In (DI) recordings of electric guitar, chopped into 0.5 second segments. The data is then processed through a SPICE model to create a "ground truth" version of the effect to train against. The training data, SPICE model, and Python code for training the networks can be found in the GainStageML/ subfolder.

License

This repository is licensed under the BSD-3-Clause license. Enjoy!

Issues· 0 open

View all issuesOpen on GitHub

No open issues yet, or sync has not completed.

> Tags

C++audio-pluginembedded-systemsguitar-pedaljuce

No comments yet. Be the first to share.

> Details

PublishedAug 1, 2026
UpdatedSep 17, 2026
Category编程语言
PricingOpen source

> Related tools

T
TypeScript
JavaScript 的超集,为前端与全栈提供静态类型
P
Python
通用编程语言,广泛用于 Web、数据与 AI
G
Go
Google 推出的简洁高效系统语言