This is a "simple", but complete example of how to utilize WebRTC to do peer to peer voice and video chatting between two or more people.
This example uses node.js and socket.io to create a "Signaling Server", which runs on (or near) your web server to manage who should talk to who. The purpose of the signaling server is to relay information between peers while you are setting them up to talk directly to each other.
Included is client.html which contains all of the logic to connect to the
signaling server, join a virtual group chat channel, connect with peers, and
stream video and audio to all party members using the raw WebRTC API.
You'll need to install node.js as well as the express and socket.io libraries:
npm install -g socket.io express
Then simply run the signaling server:
node signaling-server.js
You'll also probably want to host client.html on a web server somewhere. You'll need
to edit client.html and change YOURSERVERNAMEHERE to be the hostname or IP of
the signaling server.
Note: you can also simply open the file locally instead of using a web server, but at the time of writing this only firefox will let you access the webcam/microphone from a file hosted off of your local machine.
Now navigate to wherever you stuck client.html and you should be presented with
a dialog asking permission to access your microphone / webcam. Once accepted,
you should see a local stream appear on the page. Now open up the same page in
another browser, on the same computer or another, and watch as the magic of WebRTC takes
effect and both images and audio samples mysteriously move from one browser to the other.
Repeat with as many browsers as you dare.
Note: At the time of writing this, only firefox and chrome support WebRTC,
however both browsers support this on Windows, Linux, Mac, and Android, so lots
of fun can be had pointing everyones' phones, tablets, and laptops at that client.html
and bogging down your network with audio/video traffic.
The choice of node.js and socket.io is based purely on my familiarity with them and the fact that their fairly easy to understand even if you aren't familiar with them. However, you can use any mechanisms you want for your signaling system, you just need a way to exchange ICE candidates and session descriptions between clients.
The use of jQuery is even less important, I just like using it for DOM manipulation, and we only do that to add and remove the
You'll see client.html use adapter.js. This "library" just normalizes the
WebRTC API, which will only be necessary while WebRTC is making its way through
the standards process. Once everything is standardized and functions are
de-prefixed in the browsers, this won't be necessary anymore.