lyra is a self-hosted media server that lets you browse and watch your media with minimal fuss.

• Smart library import
• Automatically skip intros
• Synced watch sessions
• Timeline previews
• Watch state tracking
• Plex watch state import
• A lot of bugs

lyra only runs under docker for now. At some point that will change and this section will be better, but for now, here is a sample docker compose file:

services:
  lyra:
    container_name: lyra
    image: docker.io/sylver/lyra
    restart: unless-stopped
    environment:
      - PUID=1000
      - GUID=1000
    ports:
      - 8000:8000
    volumes:
      - /mnt/media/series:/mnt/media/series:ro
      - /mnt/media/movies:/mnt/media/movies:ro
      - ./.lyra:/config

There is no hardware transcoding for the time being, it adds a lot of complexity that I want to avoid while getting the basics working.

lyra uses rust + async-graphql + sqlx + sea-orm for the server and bun + react + urql for the client.

For dev you'll need rust, bun, docker, and the custom ffmpeg binaries. The easiest way to get the ffmpeg binaries is:

./scripts/export-ffmpeg.sh

That copies lyra-ffmpeg and lyra-ffprobe into ./bin, which is where debug builds look for them by default.

To start the server, from the repo root do:

cargo run

Then for the client, in client/:

bun install
bun run watch

That starts vite and proxies /api to the Rust server. In production it flips around and the Rust server serves the built client as an SPA for non-API routes.

The server will create .lyra/ locally for data and run sqlite migrations on startup.

Most of the HLS packaging logic lives in crates/lyra-packager. The workspace is split into smaller crates mostly so media-specific pieces can be tested and worked on in isolation. Most crates will have a bin for testing and development, such as lyra-timeline-preview that has a bin that dumps generated preview images for a file path.