What I meant is when the measurement would show a horizontal line (in the case of Sonarworks) this sounds the same as flat monitors in a studio with a flat signal (say white noise or music)
to the guys at Sonarworks.
What they did to determine their (lows) compensation curve is by switching between studio monitors (in a studio) and headphones.
Most likely at reference levels (say 80dB ?) and adjust EQ of the headphone so that it sounds the same as the studio speakers.
This thus also includes a bit of HRTF.
Then they measure the headphone on their (I assume home made ?) test rig which is very likely a flat-plate coupler.
The needed EQ for that headphone results in the EQ needed and overall bass compensation.
When you rinse and repeat this with several headphones you can see a pattern emerge which isolates the extra bass needed.
I have (also) used this calibration method with my rig, at least for the bass correction but not at 80dB but lower, at around 70dB as that's where I, and most other folks, listen to at longer listening sessions.
Now 80 and 70 dB doesn't sound like much difference but 80 is twice as loud. The reason why they use 80dB is obvious. Studio work.
What happens at these levels (O.W. probably also tests at these levels ?) is that the body also puts in 'information' for lower frequencies.
The higher the SPL and the lower the freq. the more the body 'listens' along.
So at 80dB you have 'added' input from the chest/nerves which is not present when listening to headphones. I believe Rin Choi called this 'the missing 6dB effect'. In any case to get the same auditory sensation when listening to music the bass needs to be turned up to get the same 'tonal balance picture'. Also O-W have this incorporated in the Harman curve because that's how people listen.
Sonarworks came up with an even higher 'correction' for the bass than O-W.
This is NOT the same as equal measured SPL with a microphones. What the plot from Sonarworks thus shows when the plot shows a 'horizontal line' (and where they EQ towards) thus is not the microphone signal but microphone signal including bass correction and some HRTF as well (for above 1kHz).
At 70dB average the 'chest thumb' effect is less taken into account by the brain and at that point a different 'correction' is needed.
When I was doing this I found the needed correction to be less than found by Sonarworks and O-W.
As most recordings are mastered at around 80dB average levels but people listening to music for longer periods to relax and enjoy I concluded that I had to take equal loudness contours into consideration.
Most likely I am not a basshead and found that when I listened at 70dB average I only needed to EQ for equal loudness contours.
When I did that the tonal balance was very similar to (nearfiled) speakers at around 70dB so my correction ended up slightly less than O-W and was not as steep/sudden as O-W & Sonarworks. Probably the FR of the body-nerves may have something to do with that and a 'threshold' for this ?
Not researched this, I don't care.
This is why the K371 measures just slightly warmer and without the (by compensation given) not existing dip at around 60Hz is not present in my measurements and is slightly north of 'neutral' to me in the lows
I hope you can see what the Sonarworks plots are showing (perceived flat to them compared to studio monitors) instead of the usual not compensated flat-plates. Most people think you take a UMIK stick it in a hole measures headphones. It doesn't, you need to calibrate those too and not using the supplied files as these are for different circumstances.
That's where my measurements are based on (not UMIK though but electret capsule that has been said to register sound very realistic.
It has drawbacks too, just like any other test rig, regardless of how expensive it is and whether or not a manufacturer bombarded it as 'reference'.
