MAGDA 0.8.0 - an open source AI-native DAW with a FAUST-powered FX bank

[lucaromagnoli]

Hi all,

I shipped MAGDA 0.8.0 today and wanted to share it here since this release is heavily built around FAUST.

MAGDA is a free, open source, cross-platform DAW (macOS / Windows / Linux, GPLv3, built on JUCE and Tracktion Engine). It includes an AI layer that operates the DAW symbolically through a small DSL: create tracks, load plugins, edit clips, build modulators, set up sidechains, etc. Every AI action becomes a normal DAW operation that is visible, editable, and undoable by the user.

Inference runs locally via llama.cpp by default, with optional API keys for larger models. The AI does not render audio directly — it operates the DAW itself.

0.8 is essentially the FAUST release: almost every built-in effect in the FX bank is now written in FAUST and compiled to native C++ at build time using the FAUST compiler (third_party/faust, currently pinned to 2.85.5). Only the Multiband dynamics and Utility devices stayed in handwritten C++.

FAUST-powered devices include:

• Filter (new): SVF, Ladder, Korg 35, Oberheim SEM, Sallen-Key, diode ladder
• Compressor, Reverb, Saturator, Clipper, Limiter, Gate
• Mod, Grit, Dimension, Pitch
• Ring Mod, EQ, Delay, Grain Delay, Chorus, Flanger, Phaser, Bitcrusher, Freq Shift

Devices with multiple engines use a single FAUST architecture selecting between sub-DSPs at compile time, exposed in the UI as an engine selector.

There is also a dedicated Faust device that lets users write .dsp code directly inside the app. It currently runs through libfaust in interpreter mode (not JIT yet), and also includes an AI panel that can generate FAUST code from a natural language prompt.

The build-time compilation workflow is what made this practical for a DAW project: the DSP stays relatively compact and declarative while still shipping as native-performance effects.

Links:

• Project: https://github.com/Conceptual-Machines/magda-core
• Download: https://magda.land
• Devices catalogue: https://magda.land/devices
• 0.8 walkthrough video: https://youtu.be/E2n1zdRG0ZA

Thanks to the GRAME team for FAUST.

Happy to answer questions about the integration or the interpreter setup.

Luca