Fast local inference
Swift services host Parakeet locally so apps can preload the model ahead of speech instead of paying a cold-start tax on the first command.
One engine to power all your voice apps.
Talk anywhere, instantly. On-device transcription that runs as a lightweight macOS daemon — one runtime for every app on your machine.
Operator view
$ vox perf --client cli
p50=132ms p95=197ms 35x realtime
$ vox transcribe /tmp/sample.wav
done: 127ms (35x realtime)
Built for products with more than one caller.
Instrumentation first
Every transcription carries stage timings and dimensions for clientId, route, and modelId so operators can inspect real latency instead of guessing.
Multi-client runtime
One daemon serves menu bar apps, browser extensions, editor integrations, and the CLI without each reinventing the runtime.
One protocol surface
Bun CLI for operator workflows, TypeScript SDK for embedded app integrations, both over local WebSocket JSON-RPC.
Measure what operators actually care about.
Stage timings for every transcription. Slice by client, route, and model.
Install once, transcribe anywhere.
Vox runs as a lightweight daemon on your Mac. Install the companion app or use the CLI — either way, any web app on your machine gets local transcription for free.
Download
Drag Vox to Applications. The daemon starts automatically on login — nothing to manage, nothing to close.
CLI
$ npx @vox/cli install
daemon installed, LaunchAgent registered
$ npx @vox/cli doctor
daemon: running
backend: parakeet
ready