[Letitroll98]

Thanks coop, very illustrative.  I'm curious what compensation (averaging is self explanatory) is used as in the raw data there's a 15db difference in the low bass and a 20db divergence in the upper treble.  Not so much in the compensated curves, just a narrow band in the high treble.  In any case the corrected plots are closer to what I hear at home, the raw data closer to what I perceived listening to the Alpha Dogs.  Funny how our perceptions change with environments.

[dB Cooper]

As I understand it, headphones tend to have uneven measured response in the top octave due to reflections and limitations of test methods- but in a way that the human ear "tunes out". This is discussed at length and from much greater understanding on the HeadRoom and Innerfidelity sites. But it is a significant issue for anyone trying to "voice" and/or measure headphones. HeadRoom et al "compensate" their curves to "cancel out" this effect in the hopes that the result will be a better indication of how the headphone sounds to th ear rather than to the test mic. "Measures flat" doesn't translate to "sounds flat", especially with material mixed/mastered on speakers. Small variations in response due to positioning variations just make the designer's job a little more complicated. "Guesstimates" must be made, one result of which is audiophiles like us having something to talk about