In Room Frequency Response Measurements


Is it better to have flat frequency response at the listening position or a smoothly rolled off treble? In your response indicate whether you listen in the near field or non-near field.
128x128onhwy61
Post removed 
My target is a gentle downward tilt across the spectrum, maybe -4 dB or so at 15 kHz relative to 400 Hz. In-room measurements are usually dominated by room acoustics at any given microphone location below 400 Hz or so, and therefore are not reliable. I mostly listen in and design for the farfield, but farfield and nearfield measurements should be very similar. Imho.

Duke
dealer/manufacturer
The reason I started this post is because I exchanged e-mails with an Audiogon member who has a very expensive system where the speaker manufacturer came to his house and spent an afternoon setting the speakers. (Yes, it's that expensive a speaker). The measured in room response at the listening position was ruler flat well into the treble. The owner says it sounds great. I have my doubts.
Post removed 
One has to wonder what they did to compensate for the natural roll off of the highs that would exist. Cone speakers droop fairly fast as the distance increases. Not so much for panels/electrostats I think. I recall many years ago trying to get FFR at the listening position with some cone speakers that were +/- 2db a 2 meters. It took an equalizer and sounded way too bright and unnatural.

What type of speakers were they?