Since you haven't spent any time learning that Audyssey is NOT an 8 band equalizer (not even close), how can you slam me???????????? Please, do some research.
Secondly, what would you propose doing about the limitations? Is there something better?
Lastly, I realize that Alan has years of experience. I also realize that Tomlinson Holman and professor Chris Kyriakakis may have a clue as well.
That's my error. Writing too quickly. I was referring to Audyssey as the whole EQ system in the receiver and not specifically the Audyssey option of the choices for EQ.
As JohnK pointed out, the manual option is limited to the parametric bands. However, the actual Audyssey option, although claiming not to use an averaging method, states they use a clustering method to "combine measurements so that acoustical problems are better represented, thus allowing the equalization filter to perform the appropriate correction at each location.".
Umm, how does a $1500 receiver manage to correct for audio 'faults' at each of 8 different locations in a room?
Clustering, such as is done in statistics to gain information about groups with seemingly large random variance within and between groups, STILL uses an averaging of sorts (comparison of variance and means) to identify clusters that are different. With that information, one can deduce where an audio difference would 'need' to be corrected with a higher probability of occurrence.
For example, one measures 5 dB down in 200 Hz at 6 of the locations, but only 5dB down in 350Hz at one location. The Audyssey would likely correct for the 200 Hz anomaly but not the one at 350 Hz.
Audyssey still uses an 'averaging' if the assumption of their clustering method is as described. In the end, it ultimately still only guesses at a single frequency adjustment curve to 'best' fit all 8 locations measured within a room.
At best it may reduce very large peaks or deep humps but again, depending on where the mic is placed those 8 times, it could worsen the sound at the seating area which is not necessarily indicative of the sound represented at the locations measured depending on the results of the clustering.
The more flat response a room at that exists to begin with, the better the Audyssey function will be at its interpretation.
The more wacky the sound is across the room, the worse its interpretation.
Well that doesn't really help people with bad rooms then does it?
I'm not claiming to make any recommendations on how to improve the sound in a room. In fact, in my media room i haven't bothered to EQ or add acoustic treatments of any kind. We put in carpet, big plush chairs, have a room that is not a perfect square or rectangle (has a ceiling with broken heights and an angled wall, with painted walls and french doors that recently had curtains installed.
It is by far the best listening room i've ever heard though i have yet to measure any frequency responses with software so perhaps this is a mental bias because i like what we did with the room. I do know i'm not 100% happy with our surround placement but with limited space, frankly i don't care. It works well enough. Of all the people who've auditioned speakers here over the past 5 years, all have agreed that the speakers sound phenomenal in the room and two have commented their Axioms didn't sound as good in their own places as they did when they were here.
Reason: room effects, obviously.
I really don't see the point of fighting a losing battle, for the time, effort and cost to eek out flattening a few decibels that most probably wouldn't even notice. I suspect there are less obvious differences to be heard than what many report without being able to easily test, blind A/B a before and after situation.
I've learned not to believe everyone or everything that one reads or 'learns' about audio even from pros (though some are more honest than others). Many of those you may think 'know' something, are really just selling another belief to sell a product whether they know it or not.
Don't just buy into the ideas but rather follow the science. Does it really make sense or not?
