Headphone Measurement Stuff

aside from the obvious problem of calibration(a mic calibrated for room measurement and calibrated for enclosed measurements aren't the same thing), I would simply try to ignore/remove compensation above 10 or 12khz. measurement above that can go wrong for so many reasons, and any small displacement might have huge impact.
the mid range has overall better stability, the higher freqs up to 10khz can move around but you usually can tell if something is wrong by ear and locate the problem rather precisely(or move the headphone a little bit in front or something). but above it becomes kind of hard to do anything precise, at least in my experience.

 

I think it's less about the coupler being "transparent" and more about it being accurate. I don't think accuracy through the upper midrange is ever going to happen with a flat plate coupler approach and perfect accuracy in the treble is never going to happen with a dummy head/in-ear-mic approach. I don't think perfect neutrality can be achieved by calibrating against the B&K target with a flat plate coupler, but it didn't sound bad when I tried it with my HD600.
I'm not sure if a completely open approach like the old JA measurements tells us all that much except if the driver is flat or not, or where it has trouble spots. Although in many cases a flat driver frequency response does sound good in headphones.

 

@ultrabike , not sure I understand what you're saying is wrong?
do you mean that moving the hd600 around doesn't impact the trebles much? because even by ear I often end up moving my 650 a little for something that annoys me around 10khz. of course my own experience is 99% IEM so there trebles are simply a nightmare to handle with positioning, plus I use a coupler that has the worst sensitivity in the trebles so after compensation I don't always know if I'm measuring the noise of something else. but I'm conscious that it's a specific problem for a specific coupler. no I was talking in general, with shorter frequencies it's easier to get into all kinds of problems, so I suggest not to rely too much on the measurements in the high freqs.

or did you talk about me saying that the calibration isn't going to be the same? if that's it then yeah you're right. it's not the calibration of the mic that changes, it's the rig and enclosure that alter the response. I use the calibration file to try and compensate for that so I do make a bad amalgam between calibration and target curve. I should have said it differently.

 

I think ringing and resonances can be suppressed or highlighted depending on the coupling material and the headphone type like you see with Marv's foam vs CD/creatology foam/felt couplers. I guess the question here is how the skin behaves vs some of the materials we have available. I might make a thread some time on some of the coupling experiments I've done with a few different materials and how they affect accuracy. LOL, I remember asking about help for my coupler on CS a long time ago and then I essentially recreated his V2 coupler and thought "WTF, this doesn't work with the HD800 at all", until I realized that the V1 coupler was very different.

What I mean is that the flat plate coupler will never be able to recreate the natural interaction between headphone and ear and since every headphone interacts differently with the ear (essentially changing the FR in different ways), the target response for a perceived neutral sound will be different for every headphone when using a flat plate coupler approach. That's why I posted all the coupler vs IE-Mic difference plots in the thread.
Actually I'm more against measurements than people realize -- I'm trying to show that there's no accurate method for measuring headphones (and also for speakers for that matter - yes, the B&K target at the listening position is no guarantee for the same or even similar FR at the eardrum. In other words different rooms/speakers need a different target response for the same perceived FR). People like to give blind trust in plots without understanding how they come about and in what way inaccuracies are involved. In what way they're compromised by design choices in the coupler design or mic choice (mainly for distortion measurements).

Of course measurements are in no way meaningless, but it's always important to show and highlight where the plots don't explain the sound. For example when the Utopia measurements had a lot of ringing in the treble, but Marv said that it subjectively sounded much smoother than measured. And then the foam coupler measurements showed clean decay with very little ringing.

@castleofargh I'm pretty sure ultrabike's coupler design is less sensitive to big changes in the FR by slight positioning differences in the upper treble. Again, this depends a lot on the coupler design but overall I would agree with ultrabike. Generally the coupler measurements are very reliable. Marv used to retake measurements from months before and they matched the old measurements almost perfectly.

 

FWIW I find my IEM results to more accurately capture tonal balance. I feel it's the other way around - we have to compensate with the coupler measurements but not with the in-ear-mic measurements. And I feel that the compensation has to be different for every headphone.

 

All of my in-ear-mic measurements are without compensation*. That plot is for speakers, not headphones. Most headphones interact very differently with the ear than speakers do with the body/head/ear, which I also showed in that thread.

*At least without compensating for the in-ear method. So are Hands' measurements. They do however have the usual mic and system calibration.

 

The simple fact is that the outer ear doesn't have as big of a gain for most headphones. Big headphones with huge drivers that are far away from the ear come closer to the interaction* you get with speakers, but they still don't show as much gain.

I meant the usual electronics calibration which Hands seems to need when I wrote system calibration. The in-ear-mic results do look like that without compensation. The area around 3-4kHz isn't boosted nearly as much as with speakers and I find flat to work well here as a target, something that I wrote here a long time ago. Actually @Tyll Hertsens' new article does seem to agree with me that speakers and headphones need different target curves at the ear drum.
The severe roll off above 10kHz with speakers depends a lot on the angle afaik. My in-ear-mic results do seem too rolled off at the top for me, but I'm not planning to compensate for that, at least not yet. I feel my coupler is more accurate in the treble anyway

I know it's hard to accept at first that the in-ear-mic results don't look like they need compensation. I wonder why noone questioned if Hands used a compensation back on CS. I know OJ was wondering about the same thing.

It's possible that the results would look very different at the eardrum, much closer to Tyll's raw plots.

*With interaction I'm simply referring to the differences in FR between no ear and with the ear, so in this case between a flat coupler and the in-ear-mic measurements.

 

That's what I'm saying. The HD600 plots with an in-ear-microphone will look mostly flat, while speaker measurements will show the huge gain. I do have speaker HRTF* measurements in my thread and the post I linked above.

*for lack of a better term. The effect the ear has on the FR for speaker measurements (against the omni mic).

Was that the ear you used for your IEM measurements? If so, try it without the ear canal. And even then I have my doubts if the fake ear is close to the real deal. The measurements Audeze showed us show us that the curves look like Tyll's results at the eardrum while they look like mine and Hands' at the ear canal opening, even with a real ear. Not really that surprising. But Audeze didn't seal off the ear canal so they got a 30db dip at 8kHz in their measurements. They might have done this on purpose to make the LCD4 measurements look better compared to the HEK.

 

Oops, getting all confused. Meant ear, not mic.

What I'm saying is there is a big difference between bypassing the ear canal and not. The difference between having to apply compensation and not.