Here is a
great article on Inner Fidelity about Dr. Sean Olive at Harmon and some of the great work he's done on headphone measurements:
http://www.innerfidelity.com/content/harman-researchers-make-important-headway-understanding-headphone-response. Thank you Tyll.
I think it's a good read and Dr. Olive's conclusions are generally similar to what I have been saying for a while...
- DF doesn't sound right in regards to headphone tuning. (Although I do feel IEMs need a slight bit of DF)
- Flat neutral response = generally good / preferred. *
- Measured performance taking into account pinna, ear canal, eardrum is highly affected by positioning, and may not be reflective of the tonal balance that we hear. (This should be obvious because speaker measurements taken by JA @ Stereophile and speaker builders have always been taken with a mic - typically a wand mic - not a dummy head.)
- After 10kHz, it's hard to get a consistent measurement, especially with a coupler. This is why I chop off my IEM measurements after 9kHz. Too much BS - I felt one too many HF'ers were taking the results past 10kHz (coupler "induced" narrow peaks) too seriously.
- And I quote: a strong preference for the speaker-in-room curves with the modified curve being most preferred. (This modified speaker-in-room curve is essentially a tilted straight line with the bass end about 10dB above the treble end.) **
*I ran a similar but much less formal and certainly not blind survey: http://www.head-fi.org/t/614631/do-objective-headphone-measurements-correlate-to-the-audiophiles-subjective-experience (study relied on experienced listeners here. it obviously wasn't taken well by a few @HF, but Dr. Olive's more formal study seems to support the initial informal survey.)
**Hmmm, where have we seen this?
http://www.changstar.com/index.php/topic,833.msg21912.html#msg21912 (minor exceptions of course. the slope downward for this set of speakers isn't quite 10db for most of the FR - probably more like 8db. Also the dip in the last half octave is a result of the mic being a few degrees off from the tweeter. But overall you should get the get the general idea. And
final tuning is always by ear.)
How are the old set of measurements on this site performed? With a sponge and then with a flat plate (the flat plate is required to get pressurization to get a good bass reading). I then combobulate the results and apply compensation curves (developed by comparing 1/3 octave noise, music, tones) to my "neutral" or "modified speaker-in-room curve" (per above definition more or less) reference. Two sets of speakers for reference. One in my living room, and one on my desk. I have also built many a speakers with "
the modified in-room curve" utilizing different drivers since I was in my teens - and that was decades ago. This stuff was never top secret and well known by studio engineers. Again, what is my methodology or my special secret? I listen, then I compare to my reference. It's that
simple. I've also relied on the ears of LFF, Anax, and to a lesser extent Ultrabike, RD, CT, Anedote, Strat, etc. and many others here as a check. And yes, the measurements have in the past been updated on a large scale as I've tweaked things, although this hasn't happened recently.
I know a few folks have stated that despite irregularities here or there, that my FR measurements tend to best to reflect what they actually hear. Well, the reason is because I actually know how "neutral" or how the "modified speaker-in-room curve" is supposed to sound like (for over 20 years now) and I've tweaked my methodology to reflect that - to use that as a reference. You can't know how this "modified speaker-in-room curve" is supposed to sound like from reading books, instruction manuals, AES papers, reading about compensation curves, mentally masturbating about whether DF or FF is more correct, etc. It's something that must be directly experienced. And no, such cannot be directly experienced at THE SHOW Newport, although two speaker systems came close this year (and I think none in 2012).
'nuff said.