Quick Start

The whole API lives at the top level of adac.

Compile a model

Given a trained FLAMO model and a sample rate, one call returns FAUST source:

import adac

faust_code = adac.flamo_to_faust(model, fs=48000, name="MyReverb")

with open("reverb.dsp", "w") as f:
    f.write(faust_code)

For inspection, the two stages are available separately. The intermediate JSON config is plain data, so it can be examined or edited between the steps:

config = adac.flamo_to_json(model, fs=48000, name="MyReverb")
faust_code = adac.json_to_faust(config, controls={"rt60": True, "dry_wet": True})

Hear it while it trains

HotReload republishes the model to a running FAUST plugin during training, so the optimisation is audible as it runs. Knob positions survive the reloads.

live = adac.HotReload(fs=48000, name="MyReverb", controls={"rt60": True})
for step in range(n_steps):
    loss = criterion(model(x), target)
    loss.backward()
    optimiser.step()
    live.update(model)
live.update(model, force=True)

Certify

certify computes a small-gain stability certificate for every feedback loop. The verdict is one of certified-stable, marginally-stable, indeterminate, not-certified, or unstable.

cert = adac.certify(config)
print(cert["verdict"])

Ship it

export_juce turns a config into an installed plugin in one call, gating on the certificate. A model whose verdict is unsafe is refused unless strict=False is passed.

adac.export_juce(
    adac.flamo_to_json(model, fs=48000, name="MyReverb"),
    "exported/", name="MyReverb",
    controls={"rt60": True, "dry_wet": True, "pre_delay": True},
    juce_modules="~/JUCE/modules",
    build=True,
)

Runnable versions of these snippets are in the examples/ directory of the repository.