Audio nerd here Study says: "Nevertheless, assuming that the perceived audio quality is largely determined by the spectral magnitude response of headphones...

Audio nerd here Study says: "Nevertheless, assuming that the perceived audio quality is largely determined by the spectral magnitude response of headphones..." This is a very wrong assumption. Audio component designers have more or less a hard time picking up which measurements can correlate with audio quality. And frequency response measurements using sine sweeps, like in the cited study, are almost of no value for discriminating between two transducers (headphones, speakers) with regarding to 'audio quality'. Also, the fact that one headphone can extend beyond 20KHz or that it can go below 20Hz will give zero guarantee of better audio quality. Frequency response measurements using white/pink noise can give a slightly better hint because they can take a look at resonant peaks that might be annoying to the listener, but even this is not a law set in stone* * Impulse measurements (and waterfall plots) can give you a clearer idea of how clear is the sound going to be; but then you can have a transducer with a fairly good impulse response but a slight resonant peak somewhere --- OR you can have sometimes a transducer which shows pretty flat frequency response but bad impulse response. A good test for intermodulation distortion (the big white elephant in the audio room) will REALLY give you a hint of which headphone will be least annoying to the ear when listening to loud complex music like classical music, vocal music, etc. It seems that the article has been written by experts in acoustics, but not really in "audio". TL;DR: Freq response measured with sine sweeps can't really tell you anything helpful to discriminate headphones with regard to sound quality.