The universal remote for your Linux multimedia server.
Part of the odio project — full documentation.
odio is an ultra-lightweight Go daemon that exposes a single clean REST API over your Linux user session's D-Bus: MPRIS players (Spotify, VLC, Firefox, MPD, Kodi), PulseAudio/PipeWire, systemd user services, and power management. No root. No hacks. Just Linux primitives.
Building a Linux multimedia setup is easy. Integrating it cleanly into Home Assistant always felt hacky, scattered integrations, SSH scripts, and fragile glue.
Tested on Fedora 43 Gnome, Debian 13 KDE, Raspbian 13, Openmediavault 8 Raspberry Pi B through Pi 5. Works without any system tweak.
# 1. Install
curl -fsSL https://apt.odio.love/key.gpg | sudo gpg --dearmor -o /usr/share/keyrings/odio.gpg
echo "deb [signed-by=/usr/share/keyrings/odio.gpg] https://apt.odio.love stable main" | sudo tee /etc/apt/sources.list.d/odio.list
sudo apt update && sudo apt install odio-api
# 2. Start
systemctl --user enable --now odio-api.service
# 3. Test (start any MPRIS player first — Spotify, VLC, MPD…)
curl http://localhost:8018/players
curl http://localhost:8018/audio/server→ See Installation for RPM, Docker, or building from source.
The built-in Odio UI is accessible at:
http://localhost:8018/ui (or http://your-host.local:8018/ui if zeroconf/mDNS is enabled)
It's a 100% local, responsive (mobile + desktop), web interface designed to control your entire Linux multimedia setup from one place: MPRIS players, per-app/global volume, systemd user services, PipeWire/PulseAudio server, and more.
There's also an installable PWA to install on your phone/desktop to easily access your remote and navigate between several instances.
odio-ha is the official Home Assistant integration for Odio.
Install via HACS → Custom Repositories → https://github.com/b0bbywan/odio-ha
What it exposes as HA entities:
media_player— global PulseAudio/PipeWire audio receiver (volume, mute)media_playerper systemd service — power on/off, volume, state tracking (MPD, Kodi, shairport-sync, etc.)- MPRIS players — auto-discovered players with full playback control and metadata (in progress)
Odio becomes the hub that makes all your HA integrations point to the correct machine. MPD service lifecycle managed by Odio, rich playback via HA's existing MPD integration — the two work together.
| Setup | What Odio gives you |
|---|---|
| RPi music server (MPD + shairport-sync) | MPRIS control + restart services from HA |
| HTPC / Kodi | Start/stop Kodi, MPRIS control via odio-ha |
| Firefox kiosk (Netflix, YouTube) | Start/stop fake Netflix and Youtube app, MPRIS control via odio-ha |
| Headless Spotify (spotifyd) | MPRIS playback + service lifecycle |
| Any PulseAudio/PipeWire setup | Per-client and global volume/mute control |
Each backend has a full reference on docs.odio.love — summaries below.
Auto-discovers all MPRIS players in real time — Spotify, VLC, Firefox, MPD, Kodi, etc. Add a player and it appears immediately, zero config. Full playback control (play/pause/stop/next/previous, seek, position, volume, shuffle, loop), real-time state via D-Bus signals, smart caching, position heartbeat. → reference
Server info and default output, global and per-client volume/mute, real-time audio events via native PulseAudio monitoring (pure Go, no libpulse). Limited PipeWire support via pipewire-pulse. → reference
Explicit whitelist required — nothing managed unless listed in config.yaml. List/monitor whitelisted system + user services, start/stop/restart/enable/disable user services, real-time state via D-Bus signals. Disabled by default. → reference
Acts as a Bluetooth audio receiver (A2DP sink) so phones and computers can stream to it — and the reverse, connecting to nearby speakers/headphones as an output (scan/connect/disconnect; devices stream live via bluetooth.discovered/bluetooth.updated SSE events). Disabled by default.
Setup needs a few system steps (Odio isn't root): add the user to the bluetooth group, install the PulseAudio/PipeWire Bluetooth module, and set Name + Class=0x240428 in /etc/bluetooth/main.conf so it advertises as an audio sink. Full guide → reference · live example on a Pi B
Remote reboot and power-off via the REST API — no SSH for day-to-day ops. Disabled by default, uses org.freedesktop.login1. On desktop, logind handles permissions automatically; on headless systems a polkit rule is required to allow the user to reboot/power-off (full rule → reference).
Agnostic upgrade frontend — Odio implements neither detection nor upgrade logic. It reads a result file written by an external detector (current/latest version, availability) and can trigger two systemd user units: one to re-check, one to run the upgrade. Capabilities are additive: the result file alone enables status reads (GET /upgrade), and each configured unit enables its trigger (POST /upgrade/check, POST /upgrade/start). Configured units are registered as internal — triggerable, but hidden from /services and the event stream. Disabled by default.
Run progress streams over a Unix socket, not a file or the journal: progress is ephemeral, and on SD-card systems a socket avoids write wear (per-user journals are also unavailable on Raspberry Pi). The upgrade script connects to the socket and writes one JSON line per milestone, relayed verbatim as upgrade.progress SSE events. The run verdict comes from the systemd job result (authoritative), independent of the script's self-report. The in-flight run state is also mirrored in GET /upgrade (under run), so a client connecting or reloading mid-upgrade still sees it.
GET /events streams live state changes to any HTTP client — no polling. One event type per backend (player.updated, audio.updated, service.updated, bluetooth.updated, power.action, upgrade.info, …), filterable with types, backend, and exclude query params (server.info is always delivered). → reference
<50msp95 response time,0%CPU on idle — tested on Raspberry Pi B and B+- Localhost binding by default, configurable per network interface
- Zeroconf/mDNS auto-discovery on the LAN (opt-in)
| Architecture | Package | Tested on |
|---|---|---|
| amd64 | deb, rpm | Fedora 43 Gnome, Debian 13 KDE |
| arm64 | deb, rpm | Raspberry Pi 3/4/5 (64-bit) |
| armv7hf | deb, rpm | Raspberry Pi 2/3 (32-bit) |
| armhf (ARMv6) | deb, rpm | Raspberry Pi B / B+ / Zero |
Pre-built packages (amd64, arm64, armv7hf, armhf/ARMv6) and a multi-arch Docker image (amd64, arm64, arm/v7) are available on every build. Docker does not target arm/v6 — Pi B/Zero users should use the armhf package.
- Wayland Remote Control, Authentication, Photos Casting...
curl -fsSL https://apt.odio.love/key.gpg | sudo gpg --dearmor -o /usr/share/keyrings/odio.gpg
echo "deb [signed-by=/usr/share/keyrings/odio.gpg] https://apt.odio.love stable main" | sudo tee /etc/apt/sources.list.d/odio.list
sudo apt update
sudo apt install odio-apiPre-built packages for amd64, arm64, armv7hf, and armhf (ARMv6) are available as artifacts on each build workflow run.
# Debian/Ubuntu/Raspberry Pi OS
sudo dpkg -i odio-api_<version>_amd64.deb
# Fedora/RHEL
sudo rpm -i odio-api-<version>.x86_64.rpmgit clone https://github.com/b0bbywan/go-odio-api.git
cd go-odio-api
task build # builds CSS + Go binary with version from git
./bin/odio-apiCreate ~/.config/systemd/user/odio-api.service:
[Unit]
Description=Dbus api for Odio
Documentation=https://github.com/b0bbywan/go-odio-api
Wants=sound.target
After=sound.target
Wants=network-online.target
After=network-online.target
[Service]
ExecStart=/usr/bin/odio-api
Restart=always
RestartSec=12
TimeoutSec=30
[Install]
WantedBy=default.targetsystemctl --user daemon-reload
systemctl --user enable odio-api.service
systemctl --user start odio-api.serviceHeadless systems: Enable lingering so the user session (PulseAudio/PipeWire, D-Bus, XDG_RUNTIME_DIR) survives without an active login:
sudo loginctl enable-linger <username>A pre-built multi-arch image is available on GHCR (amd64, arm64, arm/v7):
ghcr.io/b0bbywan/go-odio-api:latest
# 1. Prepare configuration (bind: all required for Docker)
cp share/config.yaml config.yaml
# Edit config.yaml: set bind: all
# 3. (Optional) Only needed if docker compose config shows wrong paths
cp .env.example .env
# 4. Start
docker compose up -dThe docker-compose.yml reads UID, XDG_RUNTIME_DIR, HOME and DBUS_SESSION_BUS_ADDRESS
directly from your shell environment — no configuration needed for a standard Linux setup.
See .env.example if your shell doesn't export these automatically (e.g. fish).
Environment variables passed to the container:
| Variable | Source | Purpose |
|---|---|---|
XDG_RUNTIME_DIR |
host env → fallback /run/user/$UID |
D-Bus and PulseAudio runtime directory |
DBUS_SESSION_BUS_ADDRESS |
host env → fallback derived from XDG_RUNTIME_DIR |
User D-Bus session socket |
HOME |
host env → fallback /home/odio |
PulseAudio cookie lookup path |
Volumes mounted (all read-only):
| Volume | Purpose |
|---|---|
./config.yaml |
odio configuration |
$XDG_RUNTIME_DIR/bus |
user D-Bus session socket |
$XDG_RUNTIME_DIR/systemd |
user systemd folder (utmp unavailable) |
/run/utmp |
user systemd monitoring (utmp available) |
/var/run/dbus/system_bus_socket |
system D-Bus socket |
$XDG_RUNTIME_DIR/pulse |
PulseAudio socket |
$HOME/.config/pulse/cookie |
PulseAudio cookie |
Note: bind must be set to all in config.yaml for Docker remote access (bridge network). Zeroconf won't work in bridge network mode. Host network mode is strongly discouraged.
To build locally instead:
docker build -t odio-api .
# or simply: task docker:buildThe Docker build is fully self-contained — Tailwind CSS is downloaded and compiled inside the builder stage.
--config <path>— specify a custom YAML configuration file--version— print version and exit--help— show help message
Configuration file locations (in order of precedence):
- Specified with
--config <path> ~/.config/odio-api/config.yaml(user-specific)/etc/odio-api/config.yaml(system-wide)- A default configuration is available in
share/config.yaml
Drop-in overrides (conf.d/): any *.yaml / *.yml file dropped in a conf.d/ directory next to the loaded config is merged on top of the main config in alphabetical order. Snippets override the main file, so 99-local.yaml wins over 10-base.yaml. Use this to layer per-host overrides without touching the base config — typical layout:
/etc/odio-api/
├── config.yaml
└── conf.d/
├── 10-services.yaml
└── 99-local.yaml
Disabling a backend disables the backend and all its routes.
bind: lo
logLevel: info
api:
enabled: true
port: 8018
ui:
enabled: true
sse:
enabled: true
cors:
origins: ["https://odio-pwa.vercel.app"] # default for PWA
# origins: ["https://app.example.com"] # specific originsbind: lo # loopback only (default)
# bind: enp2s0 # single LAN interface
# bind: [lo, enp2s0] # loopback + LAN (required for UI access from the network)
# bind: [lo, enp2s0, wlan0] # loopback + ethernet + wifi
# bind: all # all interfaces — 0.0.0.0 (Docker, remote access)Note: The built-in web UI requires lo to be in the bind list. If lo is absent, the UI is automatically disabled.
Each entry is a bare service name or an object {name, url} (mixable). When url is set the dashboard renders a clickable link; the shorthand :8080 resolves to the current host client-side.
systemd:
enabled: true
timeout: 90s # fsnotify stable-state timeout
system:
- bluetooth.service
user:
- mpd.service
- spotifyd.service
- name: snapclient.service
url: "http://<snapserver>:1780"
- name: mympd.service
url: ":8080"Players that need extra setup for MPRIS (MPD, shairport-sync, spotifyd, Kodi…) are covered in the systemd reference.
bluetooth:
enabled: true
powerOnStart: false # power on adapter at startup
idleTimeout: 30m # auto power-off after inactivity (0 = never)
scanTimeout: 60s # auto-stop a scan (0 = never)
power:
enabled: true
capabilities: { poweroff: true, reboot: true }
pulseaudio:
enabled: true
serve_cookie: true # exposes GET /audio/cookie for network audio clients
zeroconf:
enabled: true # mDNS (_http._tcp.local. → odio-api); disabled on `lo`
upgrade: # agnostic upgrade frontend (opt-in)
enabled: true
resultFile: /var/cache/odio/upgrades.json # required; alone it enables read-only GET /upgrade
checkUnit: odio-check-upgrade.service # optional internal user unit → enables POST /upgrade/check
upgradeUnit: odio-upgrade.service # optional internal user unit → enables POST /upgrade/start
# progressSocket default: $XDG_RUNTIME_DIR/odio-api/upgrade.sock (tmpfs, no SD writes)For upgrade, the result file is the source of truth for availability; the script reports live progress over the socket (begin/progress/end). Units are optional — omit both for a read-only status badge, or add either to enable its POST. Full per-option detail on docs.odio.love.
- Localhost binding by default — prevents accidental network exposure
- Systemd disabled by default — service control must be explicitly enabled and configured
- Read-only Docker mounts — all volume mounts are read-only in the provided
docker-compose.yml - Zeroconf opt-in — must be enabled, then mDNS adapts to
bind: disabled onlo, enabled on specific interfaces, orallinterfaces withoutlo
Full per-route reference with request bodies and responses lives on docs.odio.love/api. Overview:
| Group | Routes | Reference |
|---|---|---|
| Server | GET /server |
— |
| MPRIS | GET /players, /players/{player}/cover, POST /players/{player}/{play,pause,play_pause,stop,next,previous,seek,position,volume,loop,shuffle} |
mpris |
| PulseAudio | GET /audio, /audio/{server,clients,outputs,cookie}, POST /audio/server/{mute,volume}, /audio/{clients,outputs}/{id}/{mute,volume}, /audio/outputs/{id}/default |
pulseaudio |
| systemd | GET /services, POST /services/{scope}/{unit}/{start,stop,restart,enable,disable} |
systemd |
| Bluetooth | GET /bluetooth, /bluetooth/devices, POST /bluetooth/{power_up,power_down,pairing_mode,scan,scan/stop,connect,disconnect} |
bluetooth |
| Power | GET /power/, POST /power/{power_off,reboot} |
power |
| Upgrade | GET /upgrade, POST /upgrade/{check,start} |
below |
| SSE | GET /events |
events |
Opt-in, disabled by default — see Software Upgrades for the model.
GET /upgrade # Detector status, live run state, and available triggers (can_check/can_upgrade)
POST /upgrade/check # Re-run detection (check unit); 202 Accepted — registered only when the check unit is configured
POST /upgrade/start # Run the upgrade (upgrade unit); 202 Accepted, 409 if already running — registered only when the upgrade unit is configured
The detector's result file contract is current, latest, upgrade_available (required); checked_at is optional and must be RFC 3339 if present (dropped otherwise); anything else the detector writes (e.g. roles, manifest) is optional and passed through untouched. GET /upgrade returns the contract fields flat, the free ones grouped under extra, run while an upgrade is in flight, and can_check/can_upgrade telling clients which triggers are available:
Detector status and run lifecycle stream as upgrade.info; live run progress streams as upgrade.progress (its own type, like player.position):
// upgrade.info — detector status (on result-file change): the contract fields + "extra", no "run" or capability flags
{"current": "2026.5.0b3", "latest": "2026.6.0b1", "upgrade_available": true, "checked_at": "...", "extra": { /* ... */ }}
// upgrade.info — run lifecycle; the state is the verdict (systemd job result is authoritative):
// "idle" (none or last succeeded), "running", "failed"
{"state": "running", "percent": 42, "step": "mpd"}
{"state": "failed"}
// upgrade.progress — emitted by the upgrade script over the socket; minimum contract below, add any field you want
{"event": "begin", "total": 7}
{"event": "progress", "percent": 42, "current": 3, "step": "mpd"}
{"event": "end", "success": true}GET /events # all events (text/event-stream)
GET /events?backend=mpris,audio # filter by backend (mpris, audio, systemd, bluetooth, power, upgrade)
GET /events?types=player.updated # filter by event type
GET /events?exclude=player.position
GET /events?keepalive=60
curl -N http://localhost:8018/eventsEach line is event: <type> then data: <json>. Full event catalogue, payloads, and a browser EventSource example: events reference.
Systemd control is disabled by default and requires an explicit whitelist. Odio mitigates risks with deliberate security design:
- Disabled by default — must explicitly set
systemd.enabled: trueAND configure units. Empty config → auto-disabled even withenabled: true. - Localhost only — API binds to
loby default. Never expose to untrusted networks or the Internet. - User-only mutations — start/stop/restart/enable/disable only work on user D-Bus. System units are strictly read-only, enforced at the application layer regardless of D-Bus or polkit configuration. This protects against misconfigured or compromised D-Bus setups.
- Root forbidden by design — Odio refuses to run as root.
- No preconfigured units — nothing managed unless explicitly listed.
You must knowingly enable this at your own risk. Odio is free software and comes with no warranty.
All multimedia services run as systemd user units, not system-wide daemons. This unlocks a single, unified D-Bus session bus where PulseAudio/PipeWire, MPRIS players, and user systemd units all coexist. Odio listens to that bus and exposes everything via HTTP. Add a new MPRIS player — it appears immediately, zero code or config change.
- MPRIS Backend — D-Bus communication with media players, smart caching, real-time D-Bus signal updates
- PulseAudio Backend — native PulseAudio protocol (pure Go, no libpulse), real-time event monitoring
- Systemd Backend — D-Bus with filesystem monitoring fallback (
/run/user/{uid}/systemd/units) - Power Backend —
org.freedesktop.login1D-Bus interface
- Caching reduces D-Bus calls by ~90%
- D-Bus signal-based updates instead of polling
- Batch property retrieval
- Automatic heartbeat management for position tracking
- Connection pooling and timeout handling
- Go 1.24 or higher
go test ./...
go test -cover ./...
go test ./backend/mpris/...
go test ./backend/pulseaudio/...
go test ./backend/systemd/...The project uses Task for build automation.
# Install Task (once)
go install github.com/go-task/task/v3/cmd/task@latest
# Build for the current host (CSS + Go binary, version from git)
task build
# Cross-compile for all supported architectures (output: dist/)
task build:all-arch
# Individual targets
task build:linux-amd64 # x86_64
task build:linux-arm64 # RPi 3/4/5 64-bit
task build:linux-armv7hf # RPi 2/3 32-bit (ARMv7)
task build:linux-armhf # RPi B/B+/Zero (ARMv6, RPi OS armhf)
# CSS only
task css # Ensure CSS is available (compile or download from CDN)
task css-local # Compile locally (requires Tailwind CLI)
task css:watch # Watch mode for developmentNote: task build injects the version via -ldflags from git describe. The version is visible via ./bin/odio-api --version.
The UI uses Tailwind CSS with an intelligent multi-architecture build strategy:
- Development (x64/arm64/armv7) —
task buildcompiles CSS locally using Tailwind CLI - Legacy ARM (ARMv6 — Raspberry Pi B/B+) —
task builddownloads pre-built CSS from CDN (https://bobbywan.me/odio-css/)
Tailwind CLI doesn't provide ARMv6 binaries. The CSS is architecture-independent, so it's compiled on x64 and distributed via CDN.
CDN structure:
https://bobbywan.me/odio-css/
main/abc1234.css # commit-specific
main/latest.css # latest for branch
tags/v0.6.0.css # release tags (never cleaned)
CSS files are not committed to the repository.
Packages are built with nfpm via Task.
# Install nfpm (once)
go install github.com/goreleaser/nfpm/v2/cmd/nfpm@latest
# Build all packages for all architectures (output: dist/)
task package:all
# Individual targets
task package:deb:linux-amd64 # .deb amd64
task package:deb:linux-arm64 # .deb arm64
task package:deb:linux-armv7hf # .deb armv7hf
task package:deb:linux-armhf # .deb armhf (ARMv6, RPi OS)
task package:rpm:linux-amd64 # .rpm x86_64
task package:rpm:linux-arm64 # .rpm aarch64
task package:rpm:linux-armv7hf # .rpm armv7hl
task package:rpm:linux-armhf # .rpm armv6hl- spf13/viper — configuration
- godbus/dbus — D-Bus bindings
- coreos/go-systemd — systemd D-Bus bindings
- the-jonsey/pulseaudio — pure-Go PulseAudio native protocol (no libpulse)
- grandcat/zeroconf — mDNS / DNS-SD
- HTMX
- TailwindCSS
Odio was first pushed on January 25, 2026. It's early stage. v0.4 works out of the box, but there's a long road ahead. Expect bugs.
Does it work on your setup? What breaks? What's missing?
Try it. Tell me what works and what doesn't. Show me your setup. If you want to contribute code, even better. Go is a great language for this use case.
- Fork the repository
- Create your feature branch (
git checkout -b feature/amazing-feature) - Commit your changes (
git commit -m 'Add some amazing feature') - Push to the branch (
git push origin feature/amazing-feature) - Open a Pull Request
For issues and questions: GitHub repository
BSD 2-Clause License — see the LICENSE file for details.