What we change on Raspberry Pi OS
July 12, 2026 · View on GitHub
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.
Users & login
- The default
piuser becomesovos(uid 1000), passwordovos, member ofsudo audio pipewire rtkit ovos; autologin enabled;userconf.txtwritten so the first-boot wizard doesn't rename it back. - Lingering is enabled for
ovosso its user services start at boot with no login. - Hostname set to
raspOVOS.
System packages
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).
Python
uv+sdnotifyin system python; admin PHAL (ovos-phal,ovos-PHAL-plugin-system) in system python (root-side plugins).- The whole OVOS stack in
/home/ovos/.venvs/ovos(--system-site-packages), pinned by the ovos-releases constraints. - Skills from
skills.list(~35), STT/TTS plugins per variant/language, pre-downloaded models (wake word, intent classifiers, STT/TTS models on offline variants).
Performance & storage tuning
- zram swap via
systemd-zram-generator(compressed RAM swap instead of SD-card swap). /etc/fstabtuned byscripts/setup_fstab.sh(tmpfs for volatile paths, reduced SD writes).- sysctl tweaks for a small-memory appliance; udev rules for audio hot-plug reconfiguration.
Services
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.
What we do NOT change
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.