[Nevod]

Quote:

If an in-ear response measurement is taken with a source at 30 degrees, then the headphone is placed on-ear at 90 degrees but with the eq curve of 30 degrees, my guess is there will be a confusion be a disparity between the localization cue derived from the curve and the cues derived from the inter-aural time delay, or rather lack of it, from the headphones.  Perhaps its not a problem, or at least a minor one.  It also seems that a single point HRTF would result in a target curve that's right for that vector only, and wouldn't account for other positions in the stereo field. Couldn't find this quickly, but I know I've seen essentially polar plots of ears, and perhaps it's a flaw in memory, but I recall there is a significant difference between between response at zero, 30, and 90 degree angles.  Probably in a paper about localization.
 
I'm not convinced that the pinna isn't a factor in on-ear headphone response, but it's clearly bypassed with IEMs, so that would mean different target curves for IEM vs on-ear. Mostly at the upper end of the spectrum, of course.  No sound in real life originates from within the ear canal, so the results of that transducer position would have to be accounted for.

There will be no confusion as there is just no cues from headphones. What you will get is quite loudspeaker-like sound. Granted, the pinna changes the sound and is directional, but as apparently binaural recordings are generally reproduced quite correctly, its contribution is rather small at typical angle for headphone driver. 
You don't need many curves to emulate standard loudspeaker setup if you don't need head tracking. SVS Realiser and the like do not dissect the music into a true binaural representation as there's no information about position of source of each sound in the recording, there's just two continious channels which are intended to be reproduced by loudspeakers. What these systems do is emulating loudspeakers using HRTFs, having a set of HRTFs sor one angle would just not allow head tracking, it'd feel as if speakers are pinned to your head on long rods regardless of how you move your head.
 
A HRTF for emulation of stereo speaker setup consists of four TFs - left speaker to left ear, right speaker to right ear, left speaker to right ear, right speaker to left ear. It's easily measured and there are even some readily availible on the internet with an option to check your compatibility with it. 
There's however other nitpick. Headphones are usually designed to partially emulate diffuse-field or free-field HRTF (or some inbetween) so they could present sound at least somewhat similar to loudspeakers without using crossfeed or HRTF system. If you use an obtained real HRTF with such a HP, you'll get a mess of two HRTF overlaid. So either you'd have to use some truly flat headphones (don't know of any really, save perhaps for few Fostex mods) or equalise them back.
So, to get everything equalised, you'd have to do such steps as:
1) Equalise the speaker system. Get an TF of a setup with speakers at +-30 degrees off axis, with mic in center. If you don't have an anechoic room, to get a perfect result you'd have to get 4 TFs, just like a full set of HRTFs - because there could be some room modes present at one ear and non-present at another. Though in most cases, I think, one TF from either speaker to mic placed "inside head" would suffice.
2) Get HRTFs by using in-ear mics, having speakers compensated by inverses of TFs obtained in 1). To get really perfect you'd have to measure at eardrum, I suppose, but again, that does not seem to be practical.
3) Get a headphone TF by using in-ear mics and headphones. Again, to be perfect, it'd be good to measure at eardrum, but it's not practical - there are peaks and valleys in HP's responce which shift around with even slight change of position of a headphone, this is most pronounced in at-eardrum measurements, so such EQ would be practical only if you'd glue headphones to head. That is the part of headphone's characteristics which is not controlled by EQ and that's how real good HP differ from just good ones for one who uses EQ on headphones. 
4) Pass sound through a series of TFs - first through HRTFs of 2), then through the inverses of 3), to subtract your imperfect headphones from the chain.

 

[xnor]

Quote:

But I really don't have anything to add to armchair theories.

Then don't. This is not only theory. Such measurements have been done before, are currently done and will also be done in the future. Someone who did posted a thread in this forum (link in #37) just a day ago.
The basic procedure is also used for the Smyth Realiser - hundreds if not thousands of customers make such measurements. There are even companies that do this.

 

[bigshot]

I'm kinda more interested in how people apply theories to make their own stereos sound better than in the theories themselves. It really isn't all that different than audiophile spec sheets that keep talking about complex equasions and fractions of fractions that really don't mean much when you hit the power switch on the stereo in your living room. Where the rubber hits the road is my focus.

As long as no one gets the idea that equalization is necessarily a complicated thing because of the theorizing, it's fine. Have fun with theories.

 

[xnor]

Except for IEMs you can do exactly what we discussed before to make your headphones sound better. This has nothing to do with (fake) spec sheets.
 
What do you think happens if you start the digital room correction on your receiver, an armchair theory? Just because the consumer can push a single button doesn't mean the stuff that happens behind the curtain is trivial and what better place on this forum is there to discuss this ... other than sound science? Can your receiver do the correction also for headphones? I doubt it.

 

[bigshot]

Well maybe the problem I'm having is the complexity. You might want to make it clearer for dumb folks (like me!) If I owned IEMs and I wasn't an electrical engineer, just a normal schlub, how would I go about doing this?

 

[ComfyGrados]

Quote:

but what do you mean with the "(you'd want to modify it to be perfectly flat from 0-1000 Hz)" part?

 
I meant your target EQ curve should be flat from 0-1000 Hz since HRTF doesn't start boosting frequencies until past 1000 Hz.
 
Of course, I suppose you could always add about 6 decibels of bass to your target curve to account for the "missing 6 decibels" if you wanted to, say, tune your headphone like an ATH-M50.
 
 
Quote:

Originally Posted by xnor /img/forum/go_quote.gif
 
I don't think you need significant resources at all. See this.

 
Ha! Thanks for that link. Good to see someone else essentially doing exactly what I described.

 

[Nevod]

Proper application of HRTF has these "missing 6 dB", as HRTF for pair of speakers includes two paths to each ear - from one speaker and from another. The ear-side speaker is not changed much, the other-side speaker is delayed and attenuated above 1kHz, so below 1kHz you essentially get +6dB.
 
Here:
http://www.head-fi.org/t/612665/how-far-can-eq-really-go-towards-truly-equalizing-headphones/195#post_8948863
is a short howto from me on how to download HRTF similar to yours and apply it via stereo convolver plugin for fb2k. Seems these responces can also be used with properly configured Convolver VST, but I haven't tried that.
 
There's a drawback in these measurements though, they are not compensated for the responce of the used speaker system. On the IRCAM's site, there is a IR of their speaker system, and one can make an compensating IR. I've done that using REW, but it doesn't work with frequencies above 10kHz, so I had insane treble boost - +36dB @ 20kHz. I've used three consecutive EQs to bring that back to a reasonable level. Someone with better skills with more proper software could make an proper EQ IR.
 
After all that, bass is definitely boosted, and 'by ear', the responce corresponds to my measurements of my HPs. Be warned that most HP measurements on the web are already compensated, and most HP by themselves attempt to have a responce that roughly mimics a speaker's responce, in order not to sound too unnatural and bright. Using abovementioned setup on a normal HP, you're most likely to get too bassy and somewhat strange sound.