An honest inventory of everything the provisioning does to a stock Raspberry Pi OS Lite. If you would rather apply OVOS to an existing system yourself, use ovos-installer instead.
pi user becomes ovos (uid 1000), password ovos,
member of sudo audio pipewire rtkit ovos; autologin enabled;
userconf.txt written so the first-boot wizard doesn’t rename it back.ovos so its user services start at boot
with no login.raspOVOS.Audio stack (PipeWire, WirePlumber, ALSA utils, portaudio, mpv, ffmpeg), build tools for native wheels (build-essential, swig, python3-dev, libfann-dev, libssl-dev), camera libs (libcamera, kms++), DLNA (gstreamer, gmediarender), quality-of-life (jq, git, curl, mosh, i2c-tools, kdeconnect, zram generator).
uv + sdnotify in system python; admin PHAL (ovos-phal,
ovos-PHAL-plugin-system) in system python (root-side plugins)./home/ovos/.venvs/ovos
(--system-site-packages), pinned by the ovos-releases constraints.skills.list (~35), STT/TTS plugins per variant/language,
pre-downloaded models (wake word, intent classifiers, STT/TTS models on
offline variants).systemd-zram-generator (compressed RAM swap instead
of SD-card swap)./etc/fstab tuned by scripts/setup_fstab.sh (tmpfs for volatile
paths, reduced SD writes).See systemd units for the full unit map
— OVOS itself under the ovos user, hardware detection
(ovos-i2csound), audio autoconfiguration, splash screen, and bus
signals for reboot/shutdown as system units.
No telemetry is added anywhere. SSH stays as Raspberry Pi OS ships it.
Nothing phones home: the public STT/TTS servers are only contacted by the
lite/hybrid variants during actual use, and everything can be pointed
at your own servers — see
Change the voice or STT.