raspOVOS

August 1, 2026 ยท View on GitHub

raspOVOS turns a Raspberry Pi into a private, open-source voice assistant. You flash an SD card, boot the Pi, and talk to it. No cloud account or subscription is needed, and no data leaves your home unless you choose it.

It is the flagship OpenVoiceOS experience: ready-made images with the whole OVOS stack. Wake word, speech-to-text, intent handling, skills, and text-to-speech come preinstalled, preconfigured, and tuned for Raspberry Pi hardware.

Choose your image

Three variants trade privacy/latency against hardware requirements, each available in ~12 languages:

VariantSTTTTSNeedsBest for
litepublic serverspublic serversPi 3 might work, Pi 4 bettertrying it out on older hardware
hybridpublic serverson devicePi 4balanced daily driver
offlineon deviceon devicePi 4/5 with 4 GB+ RAM (8 GB best)full privacy, no internet dependency

!!! note "About the public servers" They are hosted by volunteers on a best-effort basis. Latency and uptime vary. If you have a bigger machine on your network, you can self-host STT/TTS and point any variant at it.

Get started in three steps

  1. Download an image from the releases page. Pick your variant and language.
  2. Flash it to an SD card. Follow Flash your first image.
  3. Boot and talk. See First boot & talking to OVOS.

Default credentials: user ovos, password ovos, hostname raspOVOS.

!!! warning Do not change the default username when flashing. The images expect the ovos user to exist.

Where to next