How to adjust EQ to create a flat response for headphones

Also, look into SonarWorks. This is a program that does exactly what you are asking for.
The problem is that it does not support your particular headphone model: Philips SHB9100.
It does support the SHB9500.
You would have to purchase another headphone that is supported by the program.
 
https://www.sonarworks.com/reference
 

rogeriodec
Thanks Mike. 
Could you tell me the difference between the blue line and the black line in the frequency response graph?

The blue graph was obtained using a special stand (SF1) for the headphone and calibrated microphones.
It has a soft surface to reduce sound conduction and reflection. It has no "ears".
EDIT: See photo here: http://reference-audio-analyzer.pro/stands.php
 
The black graph was obtained using what  is known as a "Dummy Head" stereo microphone with ears (HDM1).
The headphone is placed on the "dummy head". The ears of the dummy head contain calibrated microphone elements.  
 
That said, perhaps you should try matching the EQ curve of the black curve. It may be a better approximation of
how the headphone actually sounds.
 
And remember... you are never going to get the headphones completely flat just by looking at a chart and trying to match and invert it.  But I am sure you can improve the sound of the headphone if you try.
The biggest improvement I can see would be to take some of the low frequency boost out.
The best way to learn is to do. Set up that EQ and start turning the knobs and learn what works and what doesn't work.
 
Put the Sonitus EQ in your master buss so that the entire mix is Equalized.
Bypass the EQ before you render a stereo mix. You don't want that EQ graph printed to your mix.
 
 
 
 
 
 
 

My initial reaction to the first curve you posted was that those headphones should be fine without EQ. There are far more expensive headphones out there that don't look as good.
 
Yeh, we all hate seeing all those little hills and valleys, but in truth they are more trivial than they appear, and yours are easily within the window of what can be ear-trained away.
 
Looking at graphs, it's hard to make a distinction between significant anomalies versus inaudible variations. Valleys are less significant than peaks, and narrow valleys are often completely inaudible. But even setting aside the question of which of the bumps are important, flattening the response may still be a wild goose chase.
 
First, the assumption that "flat" is the ultimate ideal is predicated on a significant presumption: that your own hearing is flat. It isn't. 
 
Second, headphones should have a slight emphasis in the bass to compensate for the fact that they are so close to the ears. Unlike speakers that are 3-10 feet away, they don't get the acoustical assistance from the room that tends to boost bass. 
 
All that said, if I was applying EQ compensation for those cans, I'd be more concerned with the midrange dip between 1-2 KHz. Those are the most critical frequencies for mixing. I'd experiment with a modest (2-3 dB) boost there and leave the rest alone.