LSLView

Web UI for discovering and viewing Lab Streaming Layer (LSL) streams on your local network.

Because browsers cannot speak the LSL protocol directly (UDP multicast discovery + data transport), this repo includes a small backend relay that uses pylsl to read streams and forward samples to the browser via WebSocket.

Quickstart

Run the webpage and backend with this command:

curl -sSL https://raw.githubusercontent.com/onecalfman/lsl-view/main/docker-compose-release.yml | docker compose -f - up -d

It will launch on localhost:4321

Features

• Discover streams on your LAN (/api/streams)
• Stream metadata inspector (including channel labels and raw XML)
• Real-time time-series chart (Canvas, stacked/overlay, per-channel toggles)
• Event marker timeline for string streams
• Downsampling for high-rate streams

Ports

• Frontend (nginx serving built Astro): http://localhost:4321
• Backend (FastAPI): http://localhost:8765

Run With Docker

LSL discovery relies on UDP multicast. For this to work, the backend must run on the same network as the LSL sources.

Linux (Recommended)

On Linux, the backend runs with host networking so multicast discovery works:

docker compose up --build

Open http://localhost:4321, click "Scan Network", then select a stream.

Windows / macOS Docker Desktop

Docker Desktop does not reliably support the host-network + multicast setup needed for LSL discovery.

Run the backend natively, and use either the frontend container or pnpm dev.

Backend (requires Python + uv):

cd backend
uv run python server.py

Or from the repo root:

pnpm dev:backend

Frontend (Docker):

docker compose up frontend --build

Or frontend (local dev):

pnpm dev

Mock Streams (No Hardware Needed)

In a separate terminal:

cd backend
uv run python mock_streams.py

Or from the repo root:

pnpm dev:mock-streams

Standalone mock stream project:

pnpm dev:mock

This publishes:

• MockEEG (4ch @ 256 Hz)
• MockAccel (3ch @ 50 Hz)
• MockMarkers (string markers)

Then open the UI and scan.

Troubleshooting

• If "Scan Network" finds nothing: check firewall rules, that LSL sources are on the same LAN/VLAN, and that you are not trying to discover streams from inside a bridged Docker network.
• If you run the backend in Docker on Linux: verify network_mode: host is active.
• Backend health endpoint: GET /api/health.

Recording

When connected to a stream, you can start a recording from the sidebar. The backend writes:

• recordings/<timestamp>_<stream>_<id>/metadata.json – stream + backend metadata
• recordings/<timestamp>_<stream>_<id>/samples.ndjson – one JSON object per sample: { "t": <lsl_timestamp>, "d": [...] }
• recordings/<timestamp>_<stream>_<id>/recording.zip – convenience archive for download

Repo Layout

• src/ Astro + React UI
• backend/ FastAPI + pylsl relay
• docker-compose.yml production-ish setup
• docker-compose.dev.yml dev setup (hot reload)