Flat response isn't really the target for headphones. At least not a flat response in the manor these measurements are being taken.
Headphone drivers are very close to your ears which causes frequency dependent effects. Also there are issues with how the sounds bounces around your ears / ear canal / head in general (google Head Related Transfer Function for more).
A 10db drop between about 1khz and 20khz is 'about right' for a flat response during use. Additionally roll off on the low end is also expected in a measurement like this for in-ear headphones, a large amount of gain will be added once these are sealed in your ears.
I am aware of that, but I've listened to cheap earphones, I've listened to the newer low-mid-range Apple earphones, I've listened to $70-range earphones, and I've listened to mid-range ($100-200) studio headphones and mid-range ($700-1500/pr) studio monitors, and there's no way that a response curve largely matching $10 earphones (or even $70 earphones) can be described as "superb".
That said, I'm complaining about the article, not the research. By all means, Berkeley researchers, please continue developing graphene earphones!
Headphone drivers are very close to your ears which causes frequency dependent effects. Also there are issues with how the sounds bounces around your ears / ear canal / head in general (google Head Related Transfer Function for more).
A 10db drop between about 1khz and 20khz is 'about right' for a flat response during use. Additionally roll off on the low end is also expected in a measurement like this for in-ear headphones, a large amount of gain will be added once these are sealed in your ears.