adac#
Automatic differentiable audio compilation
adac compiles trained differentiable audio models to real-time DSP. It
traverses a FLAMO model graph, extracts
every parameter, serialises them to a JSON intermediate representation, and
emits valid FAUST source. From there a model reaches
a variety of hardware and software targets, plugins among them, with no manual
reimplementation.
Installation
How to install adac and the optional FLAMO and FAUST toolchains.
Quick Start
Compile a model to FAUST, hear it train, and ship it as a plugin.
API Reference
The complete top-level adac namespace: the compiler, hot-reload,
certificate, and export.
Source
The repository, issue tracker, and examples on GitHub.