Something I've investigated and wish to understand a little better in SoundID Reference.
Speakers are relatively simple in concept, flat is easy to quantify with a measurement microphone. I know there's a gradual high roll-off in SoundID which helps with translation to HiFi/Car/Consumer speakers which is standard practice.
Now headphones are very different, flat is not really flat due to the physics of people's ears and canals, accentuating high-mid frequencies and typically lowering bass. That is why Harman created their on-ear/over-ear and in-ear curves as it more closely matched peoples perception of what speakers sound like. There is a lot of information and articles online if anyone wants to research it further.
There are many other implementations of headphone EQ correction such as Oratory1990 and Crinacle who offer EQ filters to manually enter, based on the Harman 2018 on/over ear curve. However I have found the overall tone is quite different to speakers, Much of the mid-range is sucked out, the top is very bright and the bass can be fairly heavy.
I know Sonarworks have worked to make their headphone correction sound tonally the same as speakers with their correction applied, and I've spent hours testing it and comparing to other methods like mentioned above, and it does appear to be the case which is great.
I'm aware Sonarworks can't give away their IP, but having researched this myself and am very interested in what is implemented in Sonarworks speaker and headphones as a target curve? Or at least some clues to satisfy my curiosity ;-)
It is a really interesting area of research that I've spent a long time looking into, I am technical as been in IT all my life but also an amateur live and studio audio engineer, so the two things fuel my curiosity! It appears what on the surface is a simple topic, is complex, unclear and very nuanced, but very interesting!