A Steller SoundModel maker for the simple example in http://blog.teamtreehouse.com/building-a-synthesizer-with-the-web-audio-api

theremin.js

// For completeness, I include the steller API object and the current audio context as arguments.
function theremin(steller, sh) {
    var osc = null;
    var gain = sh.audioContext.createGain();
    var sm = steller.SoundModel({}, [], [gain]);

    sm.start = sh.fire(function (clock) {
        if (osc) { osc.stop(clock.t1); }
        osc = AC.createOscillator();
        osc.frequency.value = sm.frequency.value;
        osc.connect(gain);
        osc.start(clock.t1);
    });
    
    sm.stop = sh.fire(function (clock) {
        if (osc) {
            osc.stop(clock.t1);
            osc = null;
        }
    });
    
    sm.gain = steller.Param({min: 0.01, max: 1.0, value: 0.25, audioParam: gain.gain, mapping: 'log'});
    sm.frequency = steller.Param({min: 261.63, max: 493.88, value: 440.0, mapping: 'log'});
    sm.frequency.watch(function (v) {
        if (osc) {
            osc.frequency.value = v;
        }
    });
    
    return sm;
}

// Usage
var sh = new steller.Scheduler(new steller.AudioContext);
var th = theremin(steller, sh).connect(); // connect by default to the audio context's destination.
th.frequency.value = 220;
sh.play(sh.track(th.start, sh.delay(0.5), th.stop));