Privacy-first dictation: a phone app and a self-hosted server that turn speech into formatted text.
HEALTH ? 92 / Healthy
Privacy-first dictation: a phone app and a self-hosted server that turn speech into formatted text.
// readme
Open-source, privacy-first dictation system: a phone app + self-hosted server that turns speech into Whisper Flow-style formatted text, with CB-slang voice control (“Breaker Breaker” wakes, “And I’m Gone” sends).
Voice is CB, look is trucking. White/black/green, light + dark modes, CB mic motif with a digital Cobra-style LED bar meter for state. The web FE and Android app share one token set and look identical.
This repo follows a documented modular-architecture pattern — a Russian nesting-doll model. Every level is documented the same complete way:
| Artifact | File | What it’s for |
|---|---|---|
| Master architecture | ARCHITECTURE.md | The whole system: modules, connections, rules. Longer than a typical summary on purpose — it is the single source of truth. Read once; use the map + cards day-to-day. |
| Module registry (machine-checkable) | modules.toml | One entry per module: path, owns, does_not_own, depends_on, public, card. The consistency checker validates the tree against it. |
| Root module map | MODULE_MAP.md | One line per module — find the right module without opening anything. |
| Module… |