For a new project I'm working on I'd like to create an audio-reactive visualisation using analysis of the music that's playing. The new Web Audio API lets you do this pretty easily in the browser (Webkit-only at the moment).
I'm posting my documented code here with an explanation for anyone else who would like an easy snippet to get them started down the music visualisation path. You can view a demo here and download the source code here.
The main concept in the Web Audio API is that you wire up a bunch of modules together. The output from one module going into the input of another module. To this with an <audio> element, you have to create a Web Audio source with it, wire that up to an analyser, and then wire the analyser up to your speakers (so that you can hear the original <audio> again).
The code for all that looks something like this:
var audio = document.getElementById('music'); var audioContext = new webkitAudioContext(); analyser = audioContext.createAnalyser(); var source = audioContext.createMediaElementSource(audio); source.connect(analyser); analyser.connect(audioContext.destination); Once you've done that, the analyser is now listening for any output from the <audio> and will have data waiting for you whenever you ask it.
Once you set the audio.play()ing you can query the analyser at any time and check what the current frequency data of the sound is. If we put that analysis into a simple draw() loop then we can graph out what the sound looks like:
function draw() { // Setup the next frame of the drawing webkitRequestAnimationFrame(draw); // Create a new array that we can copy the frequency data into var freqByteData = new Uint8Array(analyser.frequencyBinCount); // Copy the frequency data into our new array analyser.getByteFrequencyData(freqByteData); // Clear the drawing display canvasContext.clearRect(0, 0, canvas.width, canvas.height); // For each "bucket" in the frequency data, draw a line corresponding to its magnitude for (var i = 0; i < freqByteData.length; i++) { canvasContext.fillRect(i, canvas.height - freqByteData[i], 1, canvas.height); } } You can see a demo of this running in my experiments section. I've just done a very simple line rendering of the frequency data but your imagination is the limit. Download the source code and see what you can do with it!
Stay tuned for the prettiness that I've actually got planned for this code.