Parley is an open-source chat app for communities: servers, channels, DMs, voice, video, stages and a recording studio. You run it, you keep the data, and nobody else decides the rules.
docker compose up -dText when you want to read, voice when you want to talk, a stage when you want an audience. All of it lives on one server you control.
Threads, forum channels, polls, pins, bookmarks, replies and full markdown. Mentions badge the server they happened in, and unread DMs line up under Home until you read them.
Two to ten answers, single or multiple choice, and a timer from an hour to a week. Results update live for everyone.
Peer-to-peer calls that suit rooms of about six to eight. For bigger rooms, point Parley at a LiveKit server. Push-to-talk, noise suppression and per-device settings apply live.
Speakers on stage, everyone else listening. Raise a hand to ask to speak; moderators invite you up. Schedule events with RSVPs and a reminder ten minutes before.
Record a clip from the message box. With transcription switched on, a local Whisper server writes out what was said.
"Okay, I re-amped the guitar, check the stems channel."
Coloured and gradient roles with a real hierarchy, per-channel overrides, private channels, timeouts, bans, member screening and an audit log. Custom emoji and stickers per server. List your server in Discover when you want new people to find it.
Two-factor sign-in, a device list you can log out remotely, rate limits, reports, and optional ClamAV scanning of every upload before it is stored.
A full web app plus Windows, macOS, Linux and Android apps that connect to any Parley server. Phones get a one-pane layout, and the interface speaks English, Español and Deutsch.
Nightly backups with a restore script, Prometheus metrics, JSON logs, Sentry errors, and more than one backend behind a load balancer when you grow.
A studio is a voice channel built for music. Everyone in the band records their own track in full quality, so a bad connection never ruins a take.
Engineers start and stop takes. Artists are heard and recorded. Everyone else listens from the audience until an engineer brings them in.
High-fidelity channels switch off noise suppression and echo cancellation, send stereo Opus at up to 510 kbps, and can drop to 10 ms frames for tighter timing.
Every take opens in a multitrack player with per-track gain, mute and solo, lined up automatically. Export renders a 48 kHz, 16-bit WAV of the mix.
Parley ships as two Docker images and a Postgres database. If your machine runs Docker, it runs Parley: a spare PC, a Raspberry Pi, a NAS or a VPS.
# grab the compose file and settings $ curl -fsSLO https://raw.githubusercontent.com/ParleyDev/parley/main/docs/self-hosting/compose.yaml $ curl -fsSL https://raw.githubusercontent.com/ParleyDev/parley/main/docs/self-hosting/example.env -o .env # set two secrets, then your address and admin email in .env $ sed -i "s/^POSTGRES_PASSWORD=.*/POSTGRES_PASSWORD=$(openssl rand -hex 32)/" .env $ sed -i "s/^JWT_SECRET=.*/JWT_SECRET=$(openssl rand -hex 32)/" .env $ docker compose up -d ✔ Container parley-db-1 Healthy ✔ Container parley-backend-1 Started ✔ Container parley-frontend-1 Started
The apps connect to any Parley server you choose. Type its address once and you're in.
All downloads come from the GitHub release page. The Mac build is for Apple Silicon. iOS builds from source for now.