You are ASSUMING I introduce some gear or equipment to measure the sound and create the curve, this is not so. The "curve" is entirely subjective to ME: that is, I calibrate to flat AS I HEAR IT. I do this by using frequency sweeps to identify peaks/dips, and using PEQ to correct them. I also compare random frequencies (say, 1500 Hz and 3000 Hz) to ensure they both sound equally loud TO ME in different regions). There IS no measurement gear involved, this is about the ultimate perception of the actual listener.
Here is an example - an old correction curve picture I have. This is using the Neutron PEQ. Let's say I was using my HD598 (this is old and I can't recall which HP I developed this for).
View attachment 93369
If I listen with no PEQ activated, I will hear the music as colored by the headphone/chain. Probably will sound very close to the available FR plots measured for those headphones. But if
I, (meaning only me, because this curve is the result of me compensating for MY hearing and the HP/system) activate this curve, then my perception is of a flat FR - a frequency response sweep is even in volume throughout, and coloration disappears. In this case, it is obvious I had to compensate for diminishing bass, and some treble dip and peak (countered by the graph's peak and large dip). And this is a combination of the system/HP coloration, and my OWN hearing flaws (it could be MY hearing needs that extra treble peak, and not the HP, but I am the listener and it is my perception that is corrected as well).