What is this?
WebRTC Demo is one video call app built five times: in the browser, on Android, iOS, macOS and Windows. Each one is a native app written with that platform's own tools, and they all talk to the same small servers. Any client can call any other one.
It is not a library. There is nothing to install into your project. It is a set of working apps that you can run, read and copy from.
Who it is for
- You want calls in your app and need to see how the pieces fit together on your platform.
- You are bringing WebRTC to a platform that is not here yet (Flutter, React Native, Linux, a TV) and need to know exactly what to send so the existing apps can call yours.
- You want to see how things like E2EE between a browser and a phone, iOS screen sharing in the background, or a small SFU actually work.
What you get
| Feature | Web | iOS | Android | macOS | Windows |
|---|---|---|---|---|---|
| 1:1 video call, peer to peer | ✅ | ✅ | ✅ | ✅ | ✅ |
| Group call through your own SFU (optional) | ✅ | ✅ | ✅ | ✅ | ✅ |
| Chat over a data channel | ✅ | ✅ | ✅ | ✅ | ✅ |
| Share your screen or a video file | ✅ | ✅ | ✅ | ✅ | ✅ |
| Virtual backgrounds: blur, pictures and videos | ✅ | ✅ | ✅ | ✅ | ✅ |
| Face-tracked stickers | ✅ | ✅ | ✅ | ✅ | ✅ |
| End-to-end encryption, 1:1 and group | ✅ | ✅ | ✅ | ✅ | ✅ |
| Your own microphone level on your video | ✅ | ✅ | ✅ | ✅ | ✅ |
| Picture-in-picture | ✅ | ✅¹ | ✅ | ✅² | ✅² |
¹ On iOS, picture-in-picture works in 1:1 calls only. ² An always-on-top floating window on the Mac and the compact overlay window on Windows.
What is in the repository
| Folder | What it is |
|---|---|
signaling-server/ | Node.js WebSocket relay for 1:1 calls. About 150 lines. |
sfu-server/ | Optional Go server for group calls, built on Pion. |
web/ | Vue 3 client. |
android/ | Kotlin and Jetpack Compose. |
ios/ | The iOS and macOS apps, one Xcode project. |
windows/ | WinUI 3 on .NET, with a small C shim over libwebrtc. |
effects/ | Backgrounds and stickers that every app bundles. |
tools/ | Scripts that prepare the backgrounds and draw the app icons. |
What keeps them working together
The five apps share no code across languages. What lets them call each other is a small contract that every client follows:
- the JSON messages on the WebSocket, and their order,
- who makes the offer (always the peer that was in the room first),
- the E2EE frame format and how the key is derived,
- the
media statemessage and the 300 ms camera rule, - for group calls, the SFU's protocol with two peer connections per client.
Follow it and your client can join a call with any of the five apps. Port to a new platform turns it into a checklist.
How to read these docs
- Quick start gets two clients into a call.
- How it works has one page per feature. Each page explains the idea once, then shows how every platform does it.
- Platforms shows where things live in each app's code.
- Port to a new platform is the checklist for a new client.
- Reference has every message, the versions in use and the known limits.
The same material, as one long file that also works offline, is in ARCHITECTURE.md.
A demo, not a product
The code is meant to show common WebRTC use cases and give you ideas. It may have bugs. Some things a production app needs are left out on purpose: there is no TURN server, no reconnect, no authentication, and the E2EE key goes through the server in plain form. See limitations.