...I am surprised there is as much going on between 3.5 - 4.5 khz. It was very smooth in that area despite that.
That was going to be my next question - against which standard are these measurements compensated... Harman, Diffuse Field, etc...
Again, the above frequency response measurement is uncompensated, and this is mostly how we'll be presenting frequency response measurements from now on.
Why? Because the random, willy-nilly use of compensations are a large source of confusion among those who read headphone measurements.
Many (perhaps even most) of the measurements you see online are made using
DIY headphone measurement rigs. And overwhelmingly most of the measurements I've seen posted that were made on these types of setups are compensated. But compensated using which curve(s)? I've seen some apply the diffuse field compensation based on the data from ITU-T P.58. That curve (which I think was made using a Brüel & Kjær HATS) is
not applicable to the output from
DIY headphone measurement rigs (or to those from our GRAS 45CA). I've seen some apply the Harman target (which was made using a GRAS 45CA). Again, compensation derived directly from the Harman target is not applicable to the output from DIY headphone measurement rigs. I've seen some apply compensations to the output of a specific headphone from DIY measurement rigs that were based on measurements
I've done of the same model of headphone. Again, that's not going to work either. Some simply apply a compensation curve to mold the raw output from the measurement fixture to look more reasonable when the raw output of the measurement rig would otherwise look completely wonky. And some who do these measurements are frequently changing and adjusting their compensation curves in a
season-to-taste manner.
So, in consideration of all of the above, when you look at a compensated frequency response measurement, what are you looking at? Most of the time -- when you're looking at a compensated frequency response measurement -- you literally do not know.
Over the years, there have been times I have posted diffuse-field-compensated measurements. For those,
I have used only one compensation curve, which is from diffuse field measurements made by GRAS using their KEMAR measurement manikin (which we also have here). Because there is no specific DF data for the GRAS 45CA (and given the type of fixture it is, there's not likely ever to be), I have also applied that compensation to measurements made on the GRAS 45CA (which are always indicated clearly as being diffuse-field-compensated). That said, the KEMAR DF data doesn't directly apply to measurements made on the GRAS 45CA -- and, not surprisingly, I'm less satisfied with the DF-compensated output from the GRAS 45CA using the KEMAR DF data than I am with measurements made on KEMAR. Still, when it comes to compensation, I've still only used one (again, from GRAS's KEMAR DF measurements).
And I want to be clear about something else:
You can not simply compensate one headphone measurement rig into another.
From discussions with
many engineers in trade -- most of whom you'll also only see uncompensated measurements from -- I believe posting the pre-compensated frequency response (as it came from the measurement fixture) should be the standard, and what we become accustomed to looking at. In every measurement I post, I detail the gear that was used for that measurement. I have also posted a video...
If you can't see the embedded video above, please click here.
...showing the basics of how we perform the measurement (though we have to do a minor update to that, as we've changed the procedure a bit, such that we're no longer switching to a noise stimulus before the sweep in most cases).
We'll be talking about this topic a lot more going forward.
Some key questions to ask when looking at a headphone measurement found online:
- Who did the measurement?
- What system was used to make the measurement?
- Was the headphone measurement system that was used based on standards intended to model human perception by modeling the correct impedance (of the human hearing system)?
- How are the measurements done (how is placement determined over the fixture, etc.)?
- Is the frequency response measurement shown a product of compensation?
- If the frequency response measurement was compensated, why was it compensated, and what compensation curve was used?
Right now, compensation curves are still a bit of a moving target...
As is probably clear from this post, I feel that's a vast understatement.