OfflineAudioContext FFT analysis with chrome

问题

i'm trying to build a waveform generator that get audiofile amplitudes values and display them to a canvas as quick as possible (faster than realtime) in javascript. so i use the OfflineAudioContext / webkitOfflineAudioContext , load the file and start the analyse. the waveform is to fill a wide canvas.

i analyse buffer in a processor.onaudioprocess function. (i guess it's the way it works ?)

it works fine in firefox but i've got an issue in chrome : it seems it "jumps" over much analyse to finish it works as soon as possible and only returns a few coordinates (something like 16).

here is the jsfiddle : http://jsfiddle.net/bestiole/95EBf/

// create the audio context
if (! window.OfflineAudioContext) {
    if (! window.webkitOfflineAudioContext) {
        $('#output').append('failed : no audiocontext found, change browser');
    }
    window.OfflineAudioContext = window.webkitOfflineAudioContext;
}
//var context = new AudioContext();
var length = 15241583;
var samplerate = 44100;
var fftSamples = 2048;
var waveformWidth = 1920;

var context = new OfflineAudioContext(1,length,samplerate);
var source;
var splitter;
var analyser;
var processor;
var i=0;
var average;
var max;
var coord;

processor = context.createScriptProcessor(fftSamples, 1, 1);
processor.connect(context.destination);

analyser = context.createAnalyser();
analyser.smoothingTimeConstant = 0;
analyser.fftSize = fftSamples;

source = context.createBufferSource();
splitter = context.createChannelSplitter();
source.connect(splitter);
splitter.connect(analyser,0,0);

analyser.connect(processor);

context.oncomplete = function(){
    $('#output').append('<br />complete');
}

var request = new XMLHttpRequest();
request.open('GET', "http://www.mindthepressure.org/bounce.ogg", true);
request.responseType = 'arraybuffer';
request.onload = function(){
    $('#output').append('loaded ! ');
    context.decodeAudioData(request.response, function(buffer) {
        $('#output').append('starting analysis<br />');
        processor.onaudioprocess = function(e){
            var data =  new Uint8Array(analyser.frequencyBinCount);
            analyser.getByteFrequencyData(data);
            average = getAverageVolume(data);
            max = Math.max.apply(Math, data);
            coord = Math.min(average*2,255);
            coord = Math.round((max+coord)/2);
            ctx.fillStyle=gradient;
            ctx.fillRect(i,255-coord,1,255);
            console.log(i+' -> '+coord);
            i++;
        }
        source.buffer = buffer;
        source.start(0);
        context.startRendering();
    }, onError);
}
request.send();

function onError(e) {
    $('#output').append('error, check the console');
    console.log(e);
}

function getAverageVolume(array) {
    var values = 0;
    var average;
    var length = array.length;
    for (var k = 0; k < length; k++) {
        values += array[k];
    }
    average = values / length;
    return average;
}

(here is another version that forces waveform to fit in 1920px wide and generate a waveform data for who is interested : http://jsfiddle.net/bestiole/E3rSx/ )

i really dont get the point, how doesn't chrome treat every part of the audio file ?

thanks for any help !