Local generation (default)

The main generation and audio processing workflow runs on your computer with locally installed models.

An optional remote GPU worker (phase 4, opt-in) needs a URL that implements the HTTP contract and a shared token. This repository ships a reference worker (packages/remote-worker-server). With remote disabled, no network calls are made. A remote failure does not silently start local generation.

To get started locally, install the audio engine and models by following the setup guide.