Wow! I literally haven't visited head-fi for years but this thread has got me interested again! I consider myself a "veteran equalizer", starting with matching Sony EX70's FR to that of Ety 4Ps a decade ago using graphs from Etymotic's own sales literature
but somehow all these years I never heard of this "headphone-ear resonance" phenomenon. But it all makes sense
This is my EQ for the resonances I've found for the Philips SHE3580 IEMs (pictured in avatar) I'm using in lieu of busted (again) Ety 4Ps:
This is a rough shot at an all-round EQ for the headphones to even out other FR imbalances and/or cater to my tastes:
Note that I added a -1.58dB gain within Electri-Q itself to eliminate positive gain, which should eliminate clipping. You can set a gain by adding a control point, right-clicking it and selecting Basic->Gain only.
These phones have always been very detailed with an overemphasis on the treble (I keep hearing stuff like hidden hi-hats and triangles that I never heard even with my Etys
) but I've always had trouble taming the treble without making the music sound unnatural or too dark and yet still harsh at the same time. Now I find that on many songs they sound more incredibly detailed and yet less harsh than before. Things like trailing cymbals and audience applause also seem to have acquired a unique ambience that I haven't heard before.
One note I would like to make:
It's been mentioned before, but people in this thread haven't been taking this seriously: we do not perceive sounds at different frequencies with the same loudness. Just take a look at any equal loudness contour chart, like this one from wikipedia:
You'll see that it's well documented that if you hear a sound at 20 and 2kHz as equal in loudness, they may actually be more than 50dB apart in actual SPL!!!
And this is the average reading from hundreds of subjects, the equal loudness contour for a single subject would probably be even more jagged.
Just about exactly ten years ago I wondered what music would sound like if I equalized all frequencies from my headphones to sound just as loud. And I got the same kind of curve many of you are posting, going up a cliff towards the subbass end and down a big hole around 8000Hz. And the result didn't sound remotely good.
I see a few of you writing stuff like "you have to learn that sound at high frequencies aren't louder, they are just more intense." That smells to me a lot like you did the equal loudness EQ, found it sounds like cr*p, then tried to change the EQ by ear so that it sounds more balanced on real music, then listen to the sine sweep again and tried to convince yourself that "this" is what equal loudness sounds like.
I think the only accurate way of going about this would be to
1. Do the equal loudness EQ on a reference loudspeak system, starting with a fixed reference loudness at a fixed frequency (note that equal loudness contours differ for the same person for different loudnesses!), taking care to listen to all tones from your listening sweet spot;
2. Switch to headphones with the loudspeaker-equal-loudness EQ still activated, and chain a second EQ behind that and EQ for the headphones on the second EQ, taking care to start with the same reference loudness at the same frequency on the headphones as on the loudspeakers. When you complete an equal-loudness EQ for the phones using this two-tier setup and then remove the first EQ in the chain, your headphones should have the same frequency response
at your eardrums as the loudspeak setup, assuming you were consistent in your equal loudness judgements. And to be consistent it's essential that you don't cheat yourself with thoughts like "this isn't louder, it's just more intense". Nor would you need to, since pulling out the first EQ at the end cancels out the unevenness in your equal loudness contour.
Alas, when I tried this with my loudspeaker setup I found that my non-treated room already causes strong resonances spaced less than 100Hz apart all throughout the frequency range, rendering sine sweep listening useless. Maybe when I get to visit an anechoic chamber?
For now I try to avoid turning my equalizer into an equal loudness contour graph by only looking for sharp peaks beyond the 1kHz range while letting the general upslope in loudness from 1kHz up to lie however it may after taking out the peaks. I recommend many of you to go back and try this minimal EQ again and see how you like it. I think it would be better to just EQ by ear from this point forward (admitted this is easier done on a 10-band graphic equalizer than an infinitely customizable parametric EQ...)