← back to the writings

Loading audio with the Web Audio API

Posted on September 15th, 2015 in Tutorials by Graeme

This tutorial is part of the "Seeing Sounds with Three.JS" tutorial series. In this part we will be getting started with the web audio API to add music to an existing scene.

We're building on the previous part of this series where we created a terrain. The purpose is to get that terrain moving to the beat of any chosen music track, so it looks like the floor is jumping.

However before that can be done, we first need to analyse the chosen sound file and detect the beat- and that is what we will do in this tutorial by creating an audio analyser object. The audioAnalyser will allow us to detect the beat of the music to be used in the next tutorial which will look at using that beat to move our terrain.  

The Audio Analyser

The final audioAnalyser script we're create can be seen here: 

View on Github


Overall the audioAnalyser consists of a constructor, and 5 functions:

  • constructor - sets up the objects main properties, such as the sound track.
  • loadAudio() - this does what you expect
  • decodeAudio() - once the audio is loaded, it needs to be decoded.
  • setUpAnalyser() - sets up everything needed to analyse the music
  • analyseBoost() - finds the beat
  • play() - call this to start everything off


You can use the audio analyser like this:

//create an instance (providing your sound file path)
var audioAnalyser = new AudioAnalyser('objects/audio/music.mp3');

//play music

How it works

The Audio Context

When instantiating the audioAnalyser, a number of things are set up in the constructor. The main parts are the creation of an Audio Context, the Buffer Source, and the XMLHttpRequest:

//create audio context
this.audioCtx = new (window.AudioContext || window.webkitAudioContext)();

//create buffer source
this.source = this.audioCtx.createBufferSource();       

//create XMLHttpRequest
this.request = new XMLHttpRequest();

The Audio Context is used to manage the sound passed into the constructor, and also to play it. All the functionality we need to analyse our music is provided by the Audio Context, such as decoding the music. 

A Buffer Source is then created from the Audio Context through calling audioCtx.createBufferSource(); 

The XMLHttpRequest property is used for fetching the sound from the filepath.

The Audio Buffer

Calling audioAnalyser.play() starts everything off by triggering loadAudio(), which uses the request property created in the consturctor to load our sound (as seen in line 3) :

        this.request.open('GET', this.sound, true);
        this.request.responseType = 'arraybuffer';
        var audioAnalyserInstance = this;
        //on load decode the audio
        this.request.onload = function() {
        //then send the request

Line 4 then sets a response type of arraybuffer for loading the sound. This arraybuffer then becomes part of our request object, and is used to decode the audio. That happens in line 9, where we call decodeAudio();

Decode the Audio

Here, we again use the Audio Context's (this.audioCtx) decodeAudioData function, using our request object as a parameter so that we can access the arraybuffer which it holds (line 4): 

        var audioAnalyserInstance = this;
        this.audioCtx.decodeAudioData(this.request.response, function(decodedData) {
        function(e){"Error with decoding audio data" + e.err});

The decodeAudioData also has a callback function as the second parameter (line 4), which simply calls the setUpAnalyser function of our audioAnalyser (line 6).

That's all for now

That's enough for one tutorial. In the next part, we'll look at the last section of the audioAnalyser code, and find out how an the audio context's createAnalyser function can be used to analyse the beat.

Have any thoughts?