compareeHub

BUILD A 120 DOLLARS RASPBERRY PI PEDAL THAT CLONES ANY GUITAR AMP WITH NEURAL NETWORKS

Train a neural network on your own tube amp, then run the model on a Raspberry Pi at playable latency for about 120 dollars.

by Keith Bloemer

FULL CAD BOM FIRMWARE DOCS

AudioAI

Built withRaspberry Pi

difficulty
●●●●○
time
a weekend-plus
license
GPL-3.0
repo
repo ACTIVE1,226 stars

WHAT YOU’LL NEED

Videos
1
Jump to section

COMPAREE VERDICT

This is a neural amp modeler that runs on a Raspberry Pi 4 with a HiFiBerry DAC + ADC, about 120 dollars in hardware according to the README. The concept is sound: train a neural network on audio of a real amp or pedal, then run the trained model locally with low enough latency to play live. The repository releases a ready-built plugin for Elk Audio OS, so you flash Elk onto an SD card, copy the plugin over and load models; you control model, EQ and gain from the Windows/Mac version of the plugin over WiFi. GuitarML shares many amp and pedal models, so you do not have to train your own to start. The catch is that setup still assumes comfort with command-line Linux and SSH. The repo includes a 3D-printable case for the Pi and HiFiBerry, but no footswitches or knobs - it is a box controlled over WiFi or MIDI, not a stomp pedal. Also note the README's warning: the HiFiBerry output is line level, not instrument level, so plug it into a line input, not straight into a guitar amp's front end. The payoff is a working neural modeler for the price of a mid-range commercial pedal, and the ability to capture the sound of your own amp.

GOOD TO KNOW

  • —GPL-3.0 licensed, no commercial restrictions.
  • —README lists hardware requirements and approximate 120 dollars cost to build.
  • —Code is present for both the Pi pedal and VST3 plugin builds.
  • —Setup instructions are thorough but assume you can flash an SD card and SSH into a Pi.
  • —A 3D-printable case is included; there is no wiring diagram for physical controls, which the README lists as a to-do.
  • —Community shares trained models; training your own requires audio capture of an actual amp and familiarity with Python.

Parts to buy

6 items

From our check of the build. Exact quantities and part numbers are in the creator’s BOM.

  • Raspberry Pi 4bFind
  • HiFiBerry DAC + ADCFind
  • MicroSD cardFind
  • Power supplyFind
  • Audio cablesFind
  • Optional USB MIDI controllerFind

BUILDS OF THE WEEK

Five open-source builds worth your weekend, every week.

Checked like this one: what’s really in the repo, what it costs, how hard it is. One email, unsubscribe anytime.

Can I build this?

Printoptional: the 3D-printable case, hex cover and RCA cover in the repo's 3dprint folder
BuyRaspberry Pi 4b, HiFiBerry DAC + ADC, MicroSD card, power supply, audio cables, optional USB MIDI controller
Toolssoldering iron for headers if not pre-soldered, computer to flash SD card, SSH client, optional 3D printer or enclosure fabrication tools
Skillscomfortable with command-line Linux and SSH, basic soldering if your HiFiBerry needs headers; training your own models requires Python and audio capture
Timea weekend to get it running headless; another half-day minimum if you want a case and footswitch wiring
Cost$$, dominated by the Raspberry Pi 4 and HiFiBerry board; the README estimate of 120 dollars is realistic for the core hardware
Safety5V DC only, no mains voltage. Wear hearing protection when testing with amplified guitar signals. No other hazards beyond ordinary electronics care.

Build at your own risk. Projects involve tools, electronics and sometimes mains voltage — follow the creator’s safety notes.

Videos

NeuralPi : RaspberryPi Guitar Pedal using Neural Networks

The creator's own demo video. For the build itself, GuitarML published a step-by-step guide on Towards Data Science, linked from the README.

More builds like this

All projects
A 2004 iPod streams Spotify - there's a Raspberry Pi hidden insideAudioHe built a machine that sorts loose LEGO by itself using a camera and AIAutomationThis DIY camera prints a poem instead of a photoAI

Gallery

img.youtube.com

Start here

Navigation into the creator’s own docs — we don’t rewrite the guide, we route you to the source.

  1. 1.Read the hardware requirements in the README and order the Raspberry Pi 4 and HiFiBerry board. (The README specifies the HiFiBerry DAC + ADC; Elk Audio OS must support whatever board you choose.)
  2. 2.Flash Elk Audio OS onto a MicroSD card following the setup instructions. (Elk Audio OS is a real-time Linux distribution required for low-latency audio.)
  3. 3.Download the cross-compiled Elk Audio OS plugin from the Releases page and install it on the Pi over SSH. (Assumes familiarity with SSH and command-line builds.)
  4. 4.Load a pre-trained model from the community collection or train your own. (Training your own requires Python and audio capture of the amp you want to model.)

Resources

Documentation, files and community threads for this build — we link straight to the original sources and never rehost the creator’s files.

KNOWN ISSUES

  • The README's 'around 120 dollars' covers the core hardware; budget separately for an SD card, power supply, cables and, if you print it, the case.
  • Elk Audio OS is not Raspbian; if you have never used it, expect an hour just getting SSH working and understanding the filesystem layout.
  • The repo's printed case is a tight fit (the Pi should sit flush with the four screw holes), and there are no physical knobs or footswitches - control is over WiFi or MIDI unless you add your own (the README lists physical controls as a to-do).
  • If you train your own models, use an LSTM size of 20: the README says NeuralPi is optimized for that size and other sizes are not compatible.
  • If the HiFiBerry is not seated perfectly or a solder joint is cold, you will get crackling or no audio and the error messages will not tell you why.
  • Training a model requires recording clean and distorted audio from your actual amp, which means either a loadbox or a very patient neighbour.

Can I use a Raspberry Pi 3 instead of a Pi 4?

The README is written for the Raspberry Pi 4 and the Elk Audio OS build for it; it does not mention the Pi 3 at all. Models must use an LSTM size of 20, so there is no lighter model type to fall back on. If you want to try a Pi 3, treat it as an untested experiment.

Do I need to train my own models or can I use pre-trained ones?

The community shares many pre-trained models for popular amps and pedals. You can start with those and only train your own if you want to capture a specific piece of gear.

What is the latency in practice?

The README does not give a latency figure, only that it runs models in real time on Elk Audio OS, a low-latency Linux. Try the free Win/Mac plugin first to judge the sound, and test on the Pi with your own buffer settings.

Can I run this as a plugin instead of building the hardware?

Yes, the same code compiles as a VST3 plugin for use in a DAW. The repository includes build instructions for both the Pi and plugin targets.

Community builds

No community builds yet — be the first, we feature the best ones.

Discussion1

FROM THE COMPAREE TEAM

Over 1,200 stars, a free plugin you can try on your laptop first, and a Raspberry Pi build guide — the barrier is the Linux setup and the line-level output. If you have built this or tried it, what was the one thing that took longer than the README suggested?

CompareeTEAM2mo agoedited

Practical notes from our verification: the README puts the hardware at around 120 dollars (Raspberry Pi 4, HiFiBerry DAC + ADC, Elk Audio OS), and there is a prebuilt Elk-compatible VST3 on the Releases page plus a step-by-step build guide on Towards Data Science, so you do not have to cross-compile anything yourself. The repo also ships 3D printable case parts for the Pi and the HiFiBerry board, so you are not stuck with a bare board on your pedalboard. The catch to know before you plug in: the HiFiBerry output is line level, and the README warns it should only go into line-level inputs, not straight into a guitar amp's instrument input. Development has been quiet since the last push in January 2024, so treat it as a finished project rather than an actively evolving one. If you want a packaged neural modeler, buy one; if you want to understand how they work and are comfortable with embedded Linux, this is the weekend that teaches you. Correction (4 October 2026): we re-checked this page line by line against the project's own repository, documentation and videos, and fixed errors in earlier versions.

Keith Bloemer

Keith Bloemer is the developer behind GuitarML, a collection of open-source neural network tools for guitar tone modeling. NeuralPi applies the same neural modeling techniques used in the GuitarML VST plugins to a standalone Raspberry Pi pedal, making the technology accessible outside the DAW.

GitHub

Star the project on GitHub

DISCLAIMER

  • Comparee is not the author of the projects featured here. All rights to each project belong to its creator — every page links to the original source, and we never host creators’ files.
  • Information is provided without warranty and may become outdated as projects evolve. Prices are indicative bands only — always check the creator’s parts list for current costs.
  • Building and operating any project is at your own responsibility. Protective equipment, safe workshop practice and compliance with local regulations are the builder’s responsibility.
GitHub repo