guitarix turns a Linux box into a virtual amp, a plugin or a headless rig

One engine, three ways to deliver it

guitarix is described as a modular, virtual amplifier for Linux, and what makes it interesting is that the same engine is delivered in several shapes. It comes as a standalone application, as a VST3 plugin, and as modules in the LV2 plugin format that you can drop into a DAW of your choice. It can also run headless, which the documentation points at as a way to turn a Raspberry Pi or similar device into a dedicated amp modeler, and it can be driven by a MIDI controller or a footboard. That last combination is the one with real appeal for a live rig: a headless box on the stage, a plugin on the laptop, and the same amp model in the rehearsal room. The audio layer underneath is JACK, which is stated in the project topics alongside the neural amp modeler tag, and the topics also mark the project as Linux only. Anyone expecting a cross platform installer will not find one here.

Three ways to model a preamp and a cabinet

The modelling options are the core of the project. First, there are built-in preamp and amp models you can select directly, which is the traditional amplifier simulation path. Second, you can load .nam files from the ToneHunt collection through the Neural Amp Modeler modules, which capture a specific hardware unit rather than approximating a generic circuit. Third, you can load .json or .aidax files from the Aida-X cloud collection through the RTNeural modules, which run real time neural network models. All three exist to simulate a specific hardware unit, and the difference is where the model came from and how faithful it is to that particular unit. The Aida-X and ToneHunt file collections are separate projects, so the licence terms of each model file sit outside this repository, which is worth checking before you use a capture of a commercial amplifier in a released track. A practical note for anyone trying this: a model file is only useful if the hardware it captures is the hardware you want, so the file library matters as much as the loader.

The signal chain is preamp, amplifier, effects, cabinet and impulse response

Around the amp models, guitarix provides the rest of the chain so you can build a tone rather than just load one. The documentation describes combining the preamp and amp choice with various effects and speaker cabinet emulations, and loading your own impulse response files, to arrive at a tone of your own. The effects list is long and includes compression, distortion, modulation, reverb, delay and equalisation, and some of the effect modules are explicitly modelled on popular hardware units, with the Tube Screamer named as an example based on the Ibanez unit of that name. The chain can also be extended with third party plugins, since guitarix loads LADSPA and LV2 plugins to complete the effect chain. That last point is what makes it practical inside an existing Linux DAW setup: the cabinet and amp are guitarix, and the modulation or reverb you already own plugs in alongside it. The documentation directory has separate overviews for the standalone application and for the LV2 plugins, which are the two documents to read before you start assembling a chain.

Building from source means waf, a submodule update and a long dependency list

The source build is a waf based sequence, and it assumes you want the resampler and convolver components compiled in.

git clone https://github.com/brummer10/guitarix.git
cd guitarix
git submodule update --init --recursive
cd trunk
./waf configure --prefix=/usr  --includeresampler --includeconvolver --optimization 
./waf build
sudo ./waf install

The dependency list is where the time goes. It includes gperf, intltool, sassc and python3 on the tooling side, a set of Boost packages including the iostreams and thread libraries, eigen3 for linear algebra, libsndfile, libfftw3, liblilv and liblrdf0 for the LV2 side, lv2-dev, the JACK development package in either of its two variants, GTK3 and the GTK C++ binding, and avahi plus bluetooth. On a Debian derived system that is a long apt install, and the two JACK package options are a detail worth checking against your distribution, since the names differ between the jack1 and jack2 packaging. The configure line ends with an optimisation flag, and the project also points at a container based route for people who want the build tuned to their own hardware.

GuitarixContainer compiles on the target machine instead of shipping a binary

The hardware optimisation story has its own answer in a separate project, GuitarixContainer, described as a Docker based solution. The idea is that the code is built automatically for your hardware, which is what makes low latency performance possible, especially for the Neural Amp Modeler module. The documentation frames it for Raspberry Pi devices and for laptop or workstation setups, and the linked repository is maintained separately from guitarix itself by a different author. That separation is worth noticing, because it means the container workflow is a convenience you depend on someone else's project for. For a Raspberry Pi in particular, the argument is straightforward: the Neural Amp Modeler module is a neural network inference running in real time on an ARM board, and the difference between a generic build and one compiled for that processor is the difference between playable and not. For a workstation, the same argument applies in weaker form, since a modern x86 CPU with a compiled build has more headroom than a distribution package tuned for the oldest supported processor.

Plugin formats decide how the amplifier fits into your existing setup

The delivery formats determine what the project is for you. LV2 is the open format, and guitarix's LV2 modules are the option for a DAW with LV2 support, where the amp becomes one plugin in a chain. The VST3 build lives in a separate repository, guitarix.vst, which is the path for a DAW that expects VST3 plugins. LADSPA support exists for adding effects to the chain, which is an older format and now mostly of interest for filling a gap. The standalone application is the version for playing without a host at all, and it is the one that pairs with the headless and MIDI control stories. A footnote on control: the project supports a MIDI controller or a foot-board, which means changing preset or engaging effects from a pedal without touching the keyboard. For a live rig that is the feature that decides whether the software is usable on stage, and it is the one piece of the project that the README mentions only in a single sentence.

Recent development, no stated licence, and a version scheme that tracks releases

The project is being worked on. The last push was on 2026-10-05, the repository is not archived, and the release list shows V0.47.0 in August 2025, V0.46.0 in March 2024 and V0.45.0 in February 2024. That gap between the last push and the last tag is the usual shape for a project that ships when a release is worth cutting rather than on a fixed calendar, and it means the packaged binaries you find in a distribution may be behind the source tree. The repository listing reports no licence identifier, and the README does not state one either, which is an unusual omission for a project of this size and one that has to be resolved with the maintainer before commercial use. The tree itself is small at the top level, with a documentation directory and the trunk directory that holds the waf project, plus a .gitmodules file, which is consistent with a codebase that vendors much of its dependency stack as submodules.

Editorial conclusion

guitarix suits a Linux musician who wants one amp rig that runs as a desktop application, as a plugin inside a DAW, or unattended on a small machine, and who has already collected Neural Amp Modeler captures or RTNeural models to load. It is the wrong choice for a Mac or Windows user, since the project targets Linux and the JACK audio layer, and the wrong choice for anyone who wants a supported contract, because the repository listing reports no licence identifier and the most recent tag is V0.47.0 from 2025-08-25 while the last push was on 2026-10-05. Verify three things before you commit: that the JACK server on your machine is configured for the latency you intend to play at, that the NAM or RTNeural model you want to load is available as a file you can point the module at, and whether building from source with the waf steps is acceptable to you next to the packaged distribution. The build is a waf configure and build, and the container route exists if you would rather not compile it yourself.

Frequently asked questions

Official sources

1. brummer10/guitarix on GitHub
2. Issues
3. README
4. Releases