I wonder how much that ~6k is headphone positioning variant. I don't think it would move from 6k to 10k, but a few Hz might render super-fine tunning difficult. Proly best to just do something rough that helps at different headphone positions.

In measurements, sometimes a notch will move a few Hz if you move the headphone relative to the mic.

It changes a wee bit with positioning, but it's not going to go away. Unless you wear it inside-out.

FWIW, on a Mac, you can use the CoreAudio Audio Units that come with OS X for the parametric EQ. They're supposed to be very high quality implementations.

You will need a hosting app, since iTunes cannot use it directly. I use Audio Hijack --- it's OK, though it will glitch if your computer is under load. I don't have an HD800 (yet), so I eyeballed the settings based on Tyll's published raw HD650, and applied a rough curve just to get an idea of where the EQ's headed. I was especially interested in that weird 1-3kHz rise.

I used the AUNBandEQ module with:

+4 dB low shelf at 60Hz
+4 dB parametric EQ at 1000Hz, width = 3.2
+12 dB parametric EQ at 10Hz, width = 4
63% gain (to keep the bass from clipping)

This is guesstimated stuff, and is probably way off, so consider it only as a starting point. The gain is kind of hard to set because of the AU's silly interface, but if you're below 65%, you should be OK. I heard the onset of distortion around 72%. The track I used to test for this is Bundle of Joy, the first track from the Inside Out soundtrack: listen to the deep bass notes right at 34 seconds from the end, or 2:15 of this video:



The bass is a little wooly in the YT video compared to the iTunes download, but it's OK for telling the distortion level.

Hope this useful!

Anyway, the boost in the upper-midrange just doesn't sound right to me. Sure Tyll is right in that it does bring out vocals, but it just totally kills the HD800's natural rendition of instrument and vocal timbre. Sting's voice definitely doesn't sound like that on my other headphones or any speakers I've ever built. Guitars aren't supposed to take on that plastic AKG quality. At the end of the day I can consider the possibility that I've just got used to the HD800's laid back upper-mids and don't want it to change. Either way I'd recommend a more subtle boost in that region (say 3dB), if at all.

Surprisingly the 6k adjustment isn't doing much for me. I did try playing with the Q a bit but it's actually not a dealbreaker here. Probably because the 3.0 variant of the mods are so damn good at removing that energy. The effect is there, but it's not really taking away any harshness. I was very surprised by this.

The bass boost is very welcome in my book. I've always considered the HD800 a bit bass-lite so the extra bottom does make things more palatable. I'm going to try to play a bit more by ear this time around and see what I find.

Here's my EQ-Preset for EasyQ:

<?xml version="1.0" encoding="UTF-8"?>

<Equalizer PatchFormat="2" StereoMode="Left/Right" GainRange="6 dB" GlobalGain="-1.93939394">
  <Channel>
    <Band Mode="Peak/Dip" Frequency="10641.5613" Gain="-1.93939394" Bandwidth="0.25"/>
    <Band Mode="Peak/Dip" Frequency="5336.70134" Gain="-3.27272727" Bandwidth="0.4"/>
    <Band Mode="Low Shelving" Frequency="58.1753471" Gain="2" Bandwidth="2.03"/>
    <Band Mode="Peak/Dip" Frequency="247.030078" Gain="-2.24242424" Bandwidth="1.92"/>
    <Band Mode="Peak/Dip" Frequency="3481.16203" Gain="2.39393939" Bandwidth="0.32"/>
    <Band Mode="Peak/Dip" Frequency="7173.47624" Gain="-2.06060606" Bandwidth="0.25"/>
    <Band Mode="Peak/Dip" Frequency="509.04393" Gain="-1.45454545" Bandwidth="1.36"/>
    <Band Mode="High Shelving" Frequency="14071.0216" Gain="2.54545455"
          Bandwidth="1.33"/>
  </Channel>
  <Channel>
    <Band Mode="Peak/Dip" Frequency="3424.42573" Gain="2.39393939" Bandwidth="0.31"/>
    <Band Mode="Peak/Dip" Frequency="247.030078" Gain="-2" Bandwidth="1.89996863"/>
    <Band Mode="Peak/Dip" Frequency="509.04393" Gain="-1.45454545" Bandwidth="1.73"/>
    <Band Mode="Low Shelving" Frequency="58.1753471" Gain="2.03030303"
          Bandwidth="2.08"/>
    <Band Mode="Peak/Dip" Frequency="10817.872" Gain="-2.66666667" Bandwidth="0.25"/>
    <Band Mode="Peak/Dip" Frequency="5249.72328" Gain="-1.57575758" Bandwidth="0.62"/>
    <Band Mode="High Shelving" Frequency="14071.0216" Gain="2.54545455"
          Bandwidth="1.29"/>
  </Channel>
</Equalizer>

When we developed our software getting the filters to sound good was a real pain in the rear. Implementing FIR filters isn't that hard, but developing them so they don't sound like shite is less than trivial. Our first attempts sounded like spoons being banged on old radiators - crazy pre-ringing. We were lucky to find a mathematician who knew how to code and knew a lot about audio, thus the development process was hard science driven, but always reaffirmed by ear.

Our reference curve was developed independent from Harman, but our science guys did exchange e-mails with Sean Olive to get some tips from the best. The basic process behind our curve went like this - first we developed our digital room correction tool, so we could get a flat sounding AFR from our reference speakers. Then we went on a long journey to EQ a pair of headphones, so they sound tonally close to what we got from the speakers. The idea is that one should experience very minor tonal changes when switching from neutral speakers to neutral headphones. Sure, bass and soundstage is fundamentally different on speakers and headphones, but midrange is mostly the same and that's where most of the music is.

What's interesting, initially we only had pro audio guys as our customers. We give them better mix translation, more confidence, less double checking... but they also told me that their studio speakers and headphones became more enjoyable to listen to. There is a pro-audio myth floating around that good studio speakers/headphones should sound bad. Therefore if your mix sounds good on something bad then it should sound good on everything bad or good. I don't subscribe to that and neither do my friends at Sonarworks. If your mix sounds good on a neutral system, then there's a good chance that it will be good on 80% systems out there. Also, if you need a grot-box to check your mixes, then you can always DSP change a Genelec to an NS10. Much harder to do it the other way around.

P.S. We are okay with people being cheap bastards. Just don't steal the software, our main dev is a high functioning sociopath and I can't always be around to tell him not to build in some anti-pirate (not pyrate!) tripwire.