My friend who is a much more experienced synthesist than I was telling me how important it was to consider whether the oscillators were synchronized with the gate (and if so, at what phase angle) because it affects the time-domain shape of the attack. He also said that his FFT-based spectrum analyzer showed no change even as the attack shape changed, but that you could still hear the difference.

I was doubtful because my understanding has always been that the ear can’t hear absolute phase and so that’s why, in additive synthesis, even if you randomize the phase of the harmonics, things should still sound the same. It’s particularly striking with squares and ramps; get the phase of even one harmonic wrong, and it looks like something completely different on a scope.

So I wrote a test in Pd, and while I don’t hear any differences in attack, I do hear differences in tone as the phases of the harmonics change. If you’re interested, try my test for yourself. I’m approximating a ramp using 50 harmonics, and you can control the phase of each harmonic using the phaseTable. Trigger the AR envelope with and without sync and see if you can hear a difference. Next, open up the gate so you get a steady tone and play with the phaseTable. You’ll hear the timbre change subtly, especially when changing the lower harmonics. Now I’m not sure I could tell you what phase I’m hearing in a double-blind test, but I do hear a difference. So am I wrong that the ear can’t hear absolute phase? And if I’m not wrong, why do I hear a difference? It sounds different on different speakers/headphones/earbuds so I’m guessing it has to do with asymmetries between the positive and negative excursions of the diaphragm?

additiveSaw~.pd
harmonic~.pd

@jameslo No time to check out your patch at the moment unfortunately..... maybe later..... looks interesting....

So this post might be total rubbish.

Absolute phase has some sort of meaning in analog audio because without equalisation a positive pressure at the microphone will produce a positive pressure at the speaker (unless it is wired out of phase) at exactly the same time... (not exactly true as electrons move at the speed of light...... but for us humans it holds true).
In the digital domain that cannot be true as there will be latency..... which means a phase shift..... which means the relationship no longer holds in time....... but it is just a delay that you can only hear if you are also listening to the original signal......... although the output is no longer technically in "absolute phase".

In the air it gets much more complicated..... conical speaker diaphragms...... reflections.... humidity.. pressure.....temperature...... wind...... comb filtering with stereo or if you really want to mess it up 7:1 surround..... and two ears placed where exactly......

Those elements and more produce phase differences that our brains are highly sensitive to..... its an evolutionary life saver.
So you will hear phase changes but if the phase is fixed there is no reference for your brain to access.

If you put a jack socket in your neck and feed your signal directly to your brain then you will be certain that you hear no artefacts.

So you hear changes..... but once the audio wave becomes "fixed" again in time you will "just" hear that wave.
If the gate changes the shape of the wave then I think you should be able to hear that though....

But the brain has trouble with tones because we are slow to recognise pitch..... from the Pd/doc/Manual......... cognition.zip ....... change_perception.pd is a good one.......

Probably.
I think.
David

@whale-av Hmm, so maybe I'm hearing the relative phase between partials? Seems weird though because they're just partials, all mixed together, not like the phase differences you might hear between L & R ears that help you locate sounds in space.

But I still think that I can't really tell the difference between one alignment of the AR envelope on the underlying tone over another. I've written a double-blind "lady tasting tea" test in Pd before (I compared an expensive Mogami cable with some dirty lamp cord ), but I'm not sure how to construct a test for this. Maybe you're given pairs of notes, and you have to say whether they have the same AR-to-tone phase alignment or not?