Konvoez

Konvoez is a self-hosted, single-server voice and text communication platform (Discord-like alternative where the deployed application instance acts as the server itself). Designed for friends to hang out, not for production use.

Desktop App

All versions: desktop releases. Builds are unsigned (click "More info → Run anyway" on Windows); the AppImage requires chmod +x. See the Desktop Guide.

Table of Contents

• Desktop App
• Key Features
• Deployment
• Documentation
• TODO

Key Features

• Single-Server Model: An application instance is a complete standalone server with custom text and voice rooms.
• Encrypted Messages: Text messages are encrypted server-side at rest using AES-256-GCM.
• Real-Time Voice Channels: Low-latency voice communication powered by WebRTC and Mediasoup SFU.
• Three-Tier User Roles: Built-in permission hierarchy with OWNER, ADMIN, and USER (MEMBER) roles.
• Modern Monorepo: Built with Angular (v19+, Signals) on the frontend, NestJS on the backend, and shared TypeScript schemas.

Deployment

1. Copy .env.example to .env and set required variables:

   cp .env.example .env
   nano .env

2. Start the server:

Option 1: Prebuilt Image (Recommended)

Run via docker-compose.yml:

docker compose up -d

Option 2: Building from Source

Build and run via docker-compose.build.yml:

docker compose -f docker-compose.build.yml up -d --build

Important

Initial OWNER Setup: On a fresh install, a one-time token (valid for 5 minutes) is printed to the container logs (docker compose logs). It is required to register the initial OWNER account.

Important

Voice Calls & HTTPS:

• Ports & IP: Forward UDP 40000-40100 and set MEDIASOUP_ANNOUNCED_IP in .env to your public IP/hostname.
• HTTPS Required: Browsers strictly require HTTPS for audio devices to work.
• See the Networking & Deployment Guide for reverse proxy (Caddy/Nginx) configurations.

Documentation

• Desktop Client
• Networking & Deployment Guide
• Contributing Guidelines
• Security Policy
• Password Recovery
• User Management
• Rooms
• Message attachments
• Frontend voice
• Frontend streaming

TODO

Video streaming — phase 4 (simulcast & adaptive)

• Simulcast encode layers on producers; setPreferredLayers per consumer (tile size / theatre focus)
• Bitrate caps tied to capture height; optional maxIncomingBitrate on recv transport
• Pause or reduce offscreen / hidden video consumers (adaptive downlink; align with local preview pause / theatre patterns)
• Optional viewer-side quality picker (layer selection when simulcast exists)
• Optional stream viewer count (needs signaling; not in phase 3)
• Codec auto-pick for screen (e.g. VP9/AV1) when all consumers canConsume — still no server transcode
• Mid-share height/FPS change via replaceTrack (stream quality dialog on tile / controls bar)

Other

• Swipe to reply
• Add notification click navigation
• Add common user dialog with short info direct chat/call buttons
• Add emojis or/and customizable smile/sticker packs
• Add connection state indication
• Add mute indication for other peers
• Add customizable roles (functions access) and groups (rooms access)
• Add invite emails (if email set), 2FA via email codes