I just uploaded an update for the EQ Apo config:
Harman Modification by Dreyfus Ver. 2
What I did:
- renew the pads of the reference headphone (ATH-R70X)
- remeasure the delta between the reference and the modified Nighthawk
- reduce the peaks by 4 dB (resulting in a milder equalization curve)
- add compensation curves to remove the full or half bass boost from your generic Harman equalization
You can also use the latter as an inverse filter to add missing bass induced by a lacking seal. Personally, I addressed that issue by first applying a generic
Harman equalization to my R70X, then measuring the average in-ear response, then matching the response to the Harman 2018 curve at 1 kHz, then calculating the difference between the two and adding the delta to the generic Harman equalization. You can use the arithmetic tool in REW (select A/B, then generate) to get the correction curve, then export it as text file and reformat it via batch to make it fit the GraphicEQ module in EQ Apo. You could also import the Harman curve as House Curve in REW instead and use the inbuild AutoEQ function to generate parametric settings for your headphone. However, make sure you are only applying corrections until about 1 kHz for in-ear measurements. Correcting the full bandwith would add a treble boost due to the missing ear canal resonances of your measurement. We have to rely on reference measurements (see
AutoEQ database by Jaakko Pasanen) at that point.
After the procedure described above I do now have much more confidence that the sound I am hearing is actually near the reference.
Here you can see the R70X with the generic AutoEQ correction (based on Crinacle's data) applied. I measured it with the newer flat plate (FP) and my in-ear mic (IE). As already mentioned, the flat plate has a severe downward slope due to the lack of ear resonance and overestimated seal. This is why I cannot use this data for correction (which would result in the blue curve on the bottom) without further calibration. Instead I choose the in-ear data to correct the real-life performance of the headphone worn on my head. The brown curve is the neccessary correction I have to add to the generic Harman equalization preset for the R70X (until 1 kHz).
Just for the record, here are also the newer measurements for the R70X (Harman corrected, based on Crinacle data, w/o the personal bass correction) versus my modded Nighthawk. I choose the in-ear measurements again since the flat plate ones do not correlate well with what I am actually hearing. The delta curves between the R70X and the NH are not the same for both rigs. That is why I generally question the accuracy of flat plate measurements by the way. Its correlation with my hearing is just too weak and non-linear. But that's just my take ...
The yellow curve is the actual equalization curve you can find in the
preset linked above. It is a soft Nighthawk simulation you can add on top of your Harman equalization. I hope you can find some time to try it out.
Switching back and forth I still prefer the signautre of the NH, as to be expected. The Harman curve tends to sound harsh and "in your face" overall. Especially on female voices. But we will see how I react after some longer brain in. Will report in January.
since I don't really know how to properly level-match this... thing when testing
You can easily adjust that to your liking. Just select all the data points of one specific element and shift them up and down with the arrow keys in EQ Apo until you think that it remains an average loudness when turning the module on and off.
I have tried to cover that in the second version.
In my experience that only partially works only if the 3rd listeners ear and pinna is very similar to the one generated the measurements and delta.
... its the biggest problem of headphones, namely reducing a 3D sound field to a single dimension.
Those are notable difficulties, indeed. There is obviously no easy let alone universal solution to the individuality of the HRTF topic. Also the practically mostly diffuse room acoustics with all the interferences on varying azimuths make things much more complicated.
One more reason why I like to experiment and modify standard equalization curves.
Regards
Dreyfus