Originally Posted by tomb
Meanwhile, as to the rest of it - no offense - but you make a huge assumption that the ear can't hear differences that you cite. It may seem crazily minute, but you cite no proof that the ear can't tell the difference. For example, the armatures in IEM's are almost macroscopic in transducing sound, but we hear the differences quite clearly.
Not a huge assumption, though. Broadband time delay is inaudible. If all frequencies are delayed an equal amount of time, you have no group delay issues. The only time delay becomes audible is when the time delay is non-linear over frequency. Changing the broadband time delay by 1% would be inaudible. You aren't changing the position of the transducer relative to the ear (that would possibly be audible), you aren't changing the frequencies of any possible audible resonances by enough to be audible either.
As to proof, I recognize your desire for cited references, and that's a reasonable request most of the time. However, in this case, the proof can easily be seen empirically.
Lets set the conditions first. A 10 degree F temperature change over the course of a day is, in a temperature controlled space, rather unusual, would be considered uncomfortable, but lets assume our space is only partially temperature controlled and go with a 10 degree F change. Our concern, however, is not with the ambient temperature, but with the temperature of the air as a transmission medium between the transducer and the eardrum. The air in that path is only partially influenced by the ambient temperature, as for IEMs, the ear canal is sealed, and can be assumed be be at body temperature all the time, except perhaps in the case of extremes. Our 10 degree ambient change becomes moot, because it doesn't exist in the canal. For circumnaural headphones, there is also air heating, perhaps to a lesser extent, but anyone who wears full size headphones knows they get hot. On-ear, or "open-air" headphones may possibly be the most easily influenced by ambient changes, but even in that case, the greater percentage of the air in the transmission path is within the ear canal, and is therefore heated. The balance of the air between the outer ear and the transducer will also be heated, but to a lesser extent. It would be reasonable to assume the ambient temperature change in our conditions would be reflected in that space to the extent of 20 - 30% of the total ambient change.
Now recall that if we had a transmission path length of 1.25", and that total path changed 10 degrees, we'd have a 1% broadband time delay change. If we have body-heated air, now, changing only by 30% of the total change, we now have a total time delay change of about .3%, assuming the entire transmission length changes, which it won't. I would be reasonable to assume that even with open-air headphones, a 10 degree ambient change would result in a .2% change in the transmission air path related time delay.
So, what would be audible about that? One possibility would be re-tuning of a first-order resonance, perhaps mode set up between the ear drum and the transducer. For a path length, admittedly assumed, of 1.25 inches, that mode would occur at 10,850Hz, a frequency at which we have very ability to discern frequency accurately. A .2% change would result in a retuning of that mode to 10828.3, which is a change too small to be detectable at that frequency. I'll leave the calculation of that deviation in cents to someone else.
Citing the macroscopic differences in the armatures in IEM's is irrelevant to this discussion, as that mechanism is entirely different in nature, and yes, they are clearly audible.
Now if we want to talk about how temperature affects a transducer, that's fine, might be audible, and could be tested. But again, the effect of a 10 degree ambient change is inaudible, as can be seen, again, empirically.