[MeatusMaximus]

Just wondering what anyone thinks about testing headphones on a coupler, with pinna and simulated ear canal, versus no pinna and no canal.  The second option would just be a simple plate with a mic or mics flush mounted in it.
 
The pro-pinna argument might be stated:  Real conditions include pinnae and ear canal.  Pinnae take up some of the coupling volume,  and have an effect on the response, due to reflections/resonances, particularly the concha resonance. They also have an effect on leakage (though this is highly variable for some headphones, even on real heads).  The ear canal impedance can interact with the source impedance of the headphones, much like the effect of a high output impedance amplifier (tube for example), and speakers with different input impedances.  This effect, called Pressure Division Ratio for headphones, is significant for some types of headphones.
 
The anti-pinna argument goes something like:  the pinnae you will use are "average" pinnae, so not correct for any one person.  Measuring through the ear canal adds a resonance in the measurement which "obscures" it.  The pinna and ear canal resonance are not "heard" by people, i.e. the brain takes it into account already, since we always listen through it.  So no need to measure it.
 
My argument is that the pinna measurement is more accurate.  Using "average" pinnae introduces some error, but using NO pinnae introduces more.  Including the ear canal resonance and pinna response in the measurement is not a problem.  If we want to make the measurement look visually flatter for a neutral measurement, we can use a standardized inverse ear canal filter to flatten the plotted curves, while still measuring at the ear drum (DRM reference point).
 
Thoughts?  Fire away...

 

[udauda]

My argument is that the pinna measurement is more accurate.

I back your statement 100%. No question about it. I consider the data obtained with a simple DIY jig not scientifically robust at all due to lack of repeatability too. Also thinking those coupling clamps that come with ear simulators are not really representing realistic clamping pressure/measurement angle. (though some are adjustable from 2N to 10N) Thus pinna simulation + head volume simulation are vital.

 

[MeatusMaximus]

Glad you agree, Udauda.
 
about the "clamps", we got those with the Gras 43AG Ear and Cheek Simulator.  I wondered the same thing: what are these for?  It doesn't make sense to use some particular force, even if it IS repeatable, if it doesn't represent the actual force applied by the headphones.  We use a really expensive, complicated newfangled gadget to get it right. It's called a block of wood.  It goes under the 43AG and is cut to the proper size so that the headphones are properly spaced for a typical human head, and the proper force for each set of phones is thusly applied.  I did ask our Gras rep about their "gadget" and he said it is more for holding ear buds in the concha.

 

[Nevod]

I have once posted a somewhat related question, but it got overlooked due to being obscurely named by silly me.

There were some sources studying problems of binaural sound reproduction via headphones, namely incorrect localisation. They are focused on different things, but there are key points:
1) Pressure Division Ratio varies highly with different headphone constructions. There is a criterion that is to be matched in the order to have proper localisation, and few headphones match it. Those which do match have a very acoustically transparent construction - one of them is some Stax Lambda series, which are known for their ear-to-outside transparency. Human brain is only adapted to free-air PDR, where there is no added acoustical impedance present around the ear.
2) It also varies highly on different people - the envelope of all the measurements is +- 20dB or even more (don't recall right now).
From what I've also got from these articles is that headphone FR measurements which use pinna are made at ear canal entrance, and not at the eardrum and hence don't include PDR. There is also a method that is used to measure acoustic impedance, from which PDR can be derived.

So, what do I have to say on topic? I think that only using pinna or non-pinna measurement is not enough. Both of them should be used, with some standart on non-pinna measurements, so they could be easily compared to each other. Plus, an acoustic impedance measurement is also a must to be present. From what I understand, an pinna-less measurement and acoustic impedance measurement would be enough - former would be used to determine HP's own frequency responce and later would show how it would interact with head ( not presisely, but at least the measurements could be compared to each other, plus it's also for sure that a flatter impedance means less FR distortion compared to real sound from a real source. This does not exclude pinna, as one might think, as the usual process is as follows: Headphone's FR (not individual)-> pinna transfer ratio(individual, adjusted for by brain) -> PDR produced by 2 components personal, adjusted for, and headphone-originated, not adjusted for. Hence, having flat FR and flat headphone-originated component of PDR would produce most lifelike acoustic result. ).
Though also having a pinna measurement would be more convenient, perhaps.

 

[ultrabike]

 

 

[MeatusMaximus]

I don't agree that pinna meausrements are only made at blocked meatus.  Our coupler has pinna and ear canal and eardrum simulator (711).  I know that blocked meatus measurements are recommended for some conditions, but they would not be best if the PDR was significant, since they remove the ear canal/drum impedance.
 
You could do separate measurements, but I was thinking to simplify.  If you were to only do one, would it then have pinna included?

 

[udauda]

1) Pressure Division Ratio varies highly with different headphone constructions. There is a criterion that is to be matched in the order to have proper localisation, and few headphones match it. Those which do match have a very acoustically transparent construction - one of them is some Stax Lambda series, which are known for their ear-to-outside transparency. Human brain is only adapted to free-air PDR, where there is no added acoustical impedance present around the ear.
2) It also varies highly on different people - the envelope of all the measurements is +- 20dB or even more (don't recall right now).
From what I've also got from these articles is that headphone FR measurements which use pinna are made at ear canal entrance, and not at the eardrum and hence don't include PDR. There is also a method that is used to measure acoustic impedance, from which PDR can be derived.

1) PDR is not much of a big deal with supra-aural, circum-aural, and supra-concha headphone, according to [1,2]. It's only a few dB difference with narrow-bandwidth! Your assumption would only be applicable when the ear(overall) acoustic input impedance is interfered. Such case would be with insert earphones and intra-concha headphone. Simply by using a IEC 60318-4 simulator shall effectively eliminate the trouble.
2) According to [3], such interpersonal deviation much lessens when the ear canal is omitted(blocked-ear canal), and substituted with a standard transmission correction figure from an occluded ear simulator to ear canal extension. Conventional HATS/Ear-n-Cheek simulator measurements, including Tyll's and mine, almost always include the PDR so I'd leave the issue at rest.
 
I've no problem with current international standards related to headphone measurement, but hope they increase the frequency range of IEC 60318-4 further up to 20 kHz. 

 Otherwise the data measured with an occluded ear simulator must have the same limit. Dr.Bruel (B&K) and Dr.Rasmussen (GRAS) did the data verification of their new ear coupler together in 1976, and the final in-ear impedance target is still robust even today.
 
[1] H. Møller, D. Hammershøi, C. B. Jensen, and M. F. Sørensen, "Transfer characteristics of headphones measured on human ears," J. Audio Eng. Soc. Vol. 43, (April 1995).
 
[2] D. Hammershøi, H. Møller, "Sound transmission to and within the ear canal," J. Acoust. Soc. Am. Vol. 100, (1996).
 
[3] D. Hammershøi, H. Møller, "Binaural Technique - Basic Methods for Recording, Synthesis, and reproduction", Communication Acoustics, J. Blauert (Ed.), Springer: Berlin (2005).

 

[Nevod]

MeatusMaximus
I don't say that there's no measurements at eardrum, just that it was stated they're uncommon, not even by me.
 
udauda
Unfortunately, I do not have access to any of these papers and only things I've seen are citing them. 
Namely is that one: http://www2.ak.tu-berlin.de/~akgroup/ak_pub/2011/Schultz_2011_Anextraauralheadphoneforoptimizedbinauralreproduction_26TMT.pdf
It's an attempt to make an headphone for binaural reproduction, and the paper refers to your [1] paper. It states that there is an critertion to be met for proper reproduction, called FEC(Free-air Equivalent Criterion), and that 'phones compliant with that criterion have to have PDR kept within some limit, it is not directly stated, what the limit is, but seems like it's ±6dB, and it is stated that their experimental phone matches it like 2 other extraaural phones in the Moller's paper.
 
I unfortunately can't find the second paper I've read, but I recall that it's related to your 2nd point: the replacement of an ear canal with a model is not a problem by itself as we're not measuring the head, the problem is that PDR changes with different headphones, so we can't really use one PDR compensation curve for all, we have to measure it and compensate then.
 
However, as I've stated, I don't have access to the source papers and can't really draw conclusions here.
 
As for measurements, again, I think I'd have to look at the recent AES paper discussed in the next thread to reliably say anything. However, as a DIY person, I'd actually like to have some simple and repeatable standard for measurement that would allow interchanging measurements and reliably comparing measurement of one person to another. That would probably preclude using pinna for measurement. There however should be some, probably centralized, comparison of no-pinna measurement done by that standard with an with-pinna measurement, so it would be possible to apply a compensation and have an with-pinna measurement "at home". Though my common sense tells me it won't likely be correct due to various reflections present with pinna and much less reflections present with a baffle, but without serious dive into the topic I don't know.

 

[MeatusMaximus]

I'm not sure I'd agree that PDR is not a problem with circumaural headphones.  True, the data shown in the original Muller/Hammershoi papers does seem to show modest variations, though in the "Transfer Characteristsics..." paper they only show up to 7 kHz (and PDF is increasing with freq. up to that point).  Also of course it depends on the headphones.  Dorte Hammershoi was one of my profs at Aalborg, and she has repeatedly reminded me that just because a headphone is "open" does not make it FEC, and vise versa.  Hard to know which headphones will cause problems.
 
Florian Volk at TU Munich has also done some measurements of PDR and is measuring a bunch more headphones as part of a project.  it will be interesting to see what he comes up with.
 
Nevod, would I get in trouble if I send you one or two of the AES papers?  I hate to see you not able to read them!  OR, you can buy them pretty cheap on the AES site.

 

[udauda]

Quote:

I'm not sure I'd agree that PDR is not a problem with circumaural headphones.

I think it's little off from the topic, since DRP measurements with pinna simulation backed by us reflect the impedance interaction between the ear and the headphone driver. Anyway, that is the general agreement elicited between those two researchers, Dr.Hammershøi and Dr.Møller, in 2008, not by me. Shoot her an email and you will see why. Above 7 kHz, according to Hammershøi, the response deviation is due to highly individual differences in ear canal geometry. I can't possibly think the PDR above 7 kHz can be generalized.

 

[MeatusMaximus]

So, you are saying that the effect is small, but would be accounted for anyway  with the coupler measurement.   I'm just saying the effect might be significant for some headphones, but agree that it should be accounted for with the coupler measurement.  And yes, I agree that the blocked meatus measurement has been shown to have less variability.  Still, that is mostly in the context of individual measurements (with all the problems that arise from trying to stick a tube in someone's resonant ear canals in a repeatable way).  For measuring with a coupler, it should be more consistent.  
 
I would be curious to know what you think about the leakage problem.  It is not to hard to measure headphones with zero leakage around the cup, using a flat plate and/or pushing on the phones slightly.  however, real cups do leak significantly with some phones, depending greatly on exact placement and even the individual wearing them.  So, how do you measure it?  I've seen differences of easily 20 dB at the low end with and without leak.  I doubt that there is a satisfying answer to this problem...

 

[udauda]

So, you are saying that the effect is small, but would be accounted for anyway  with the coupler measurement. I'm just saying the effect might be significant for some headphones, but agree that it should be accounted for with the coupler measurement.  And yes, I agree that the blocked meatus measurement has been shown to have less variability.  Still, that is mostly in the context of individual measurements (with all the problems that arise from trying to stick a tube in someone's resonant ear canals in a repeatable way).  For measuring with a coupler, it should be more consistent.

Yes, sir. I agree.
 

I would be curious to know what you think about the leakage problem.  It is not to hard to measure headphones with zero leakage around the cup, using a flat plate and/or pushing on the phones slightly.  however, real cups do leak significantly with some phones, depending greatly on exact placement and even the individual wearing them.  So, how do you measure it?  I've seen differences of easily 20 dB at the low end with and without leak.  I doubt that there is a satisfying answer to this problem...

I believe Plantronics, Inc. in California demonstrated real-ear leakage vs simulator(B&K 4128C with Type 3.3 pinna) leakage comparison at the 114th AES convetion in 2003, and the difference between the ear and the simulator was more or less the same with intra-concha, supra-concha, and supra-aural headphones(open & closed pad). Considering how round B&K 4128C is, it seems HATS data indeed reflect realistic amount of acoustic leakage. Which means, using a flat-plate is not really ideal.
 
IMHO though, I would just measure under zero leakage with a flat plate 

, and that is how I designed my dummy head as well. Sub-bass loss & distortion caused by leak is just too much of hassle to bother about! However, under a circumstance, in which realistic simulation of leakage is vital in order to estimate subjective preference from simulator data, I think a HATS is an absolute necessity.

 

[MeatusMaximus]

Now that you mention it, Gras has some ear simulators that have controlled leaks.  I'm at home now so don't remember the model.  But it does not have pinnae.  Not sure what it is used for, perhaps telephonic measurements.
 
I'll have to check into the Plantronics thingy.  So, you are thinking that the rounded head shape and type 3.3 pinnae on the HATS would give a fair assessment of leakage.  That sounds as reasonable as anything else I have heard or thought of.  But still, on some phones the leakage will vary a lot just by placement.  So, perhaps multiple re-seats and averaging.  We are also measuring  at blocked canal on human subjects, and that is pretty realistic for leakage, if you can get a reasonable number of subjects measured.  Lot of work though.

 

[xnor]

One more anti-pinna argument could be CSD visualization. Some people seem to have got the (wrong) idea that a headphone needs to have such a fast decay as a tweeter tested in an anechoic chamber.

 

[MeatusMaximus]

Quote:

One more anti-pinna argument could be CSD visualization. Some people seem to have got the (wrong) idea that a headphone needs to have such a fast decay as a tweeter tested in an anechoic chamber.

I personally don't tend to think in terms of CSD or related depictions, hard for me to interpret.    But even so, do you think pinna would make the decay longer?