I've just registered with this very impressive forum.
I have a Denon 2807, and before retirement had some professional involvement with motion picture sound recording and reproduction techniques. In reading the material on the web in general, as well as this forum, I find very little about what Audyssey actually does in terms of transducer and room equalization.
I'm clear that it normalizes time delays of the several speakers, does something about equalization, and in some way averages the responses in a number of different listening locations. Whatever it does requires little input from the user, just push the buttons and move the microphone around.
I have no real understanding about the nuts and bolts of Audyssey's spatial averaging or equalization. In the 2708 system,the manual says the first position is given extra weight, and should be the preferred listing position, but provides no further details. Something is said about using only direct path response for the measurements, but not how this is done. The only way I can envision is some kind of time gating, but I don't know know to implement this with the FM sweep test chirp used by my receiver. In any event, I would not think it possible to optimally correct all positions at the same time.
I have some concern with the emphasis placed on wide-band (noise or FM chirp) analysys. Frequency response is by nature narrow band. You want to know what the system does with a high degree of resolution. A narrow band or high Q resonance takes time to be excited and to decay. For this reason it is less obvious with wide band measurements, but nonetheless real. It's important to view these things correctly. A narrow resonance is not as easily heard with music or speech as is a broader resonance. Perception seems more proportional to the area under the curve. ("see Audibility of Resonances" chapter at: http://www.infinitysystems.com/home/...rt_science.pdf
There are several interesting papers concerning loudspeakers and room acoustics here.
The Audyssey web site (www.audyssey.com/technology/graphs/graph1.htm
shows a series of graphs, and states the response is measured at "hundreds of points" and a complementary equalization curve generated.
Well, I counted all the points in the graph they showed.and I got 173, probably +/- 10. 77 points between 30 and 150 Hz, about 90 between 150 and 10KHz, and 6 between 10 and 20 KHz. I've got to believe that this, if it represents an actual system, it is the very best one available. It is not said if these points represent the sum of all possible measurement points, or if measurements are in fact made at each point, and the results saved for further computation.
Another page (www.audyssey.com/technology/multEQ_products.html
) shows a chart with comparisons between 3 versions of Audyssey. There is a little confusion here also. The 3 versions are MultEQ-XT, MultEQ and 2EQ. Denon shows the 2807 as using MultiEQ-XT, but that is contradicted as 6 measurement positions are provided in the 2807 but the site text specifies that in the non-professional, reciever version of -XT there are 8 measurement positions. It is the stand-alone device, requiring professional installation that has the highest performance shown on this site. This suggests that the 2807 version more resembles what the site describedsas MultEQ.
If I use the comparisons in the Audyssey chart, for a MultEQ version I get about 10 measurement points "for the subwoofer" for which I am assuming a 150Hz crossover. This corresponds with the counted 77 points shown in the graph for the full-bore version. The rest of the spectrum gets 16 measurement points. This is roughly comparable to a 1/3 octave equalizer, and some measurements of before and after and compensation curves published on this site tend to support that kind of resolution performance. I don't remember seeing any information about the magnitude of boost or cut available, but the published graphs suggests it is cut only.
Now, Audyssey says it is neither a constant or variable Q graphic, or a parametric equalizer, but nowhere can I find what it is. It seems implied that the implementation is of the weighted tapped delay type, and is linear phase (constant delay) in its time domain response. I can find no hard data about the resolution, spacing or number of measurement or compensation points, especially in Audyssey's simpler versions.
Possibly this lack of information is due to business reasons, but it is not obvious what they are. Other traditional equalizers are available in linear phase form, and their pros and cons are pretty well understood. I know I would feel I could better use and evaluate Audyssey if I had engineering data to supplement the almost entirely subjective opinions I now have. The nearly automatic operation is certainly helpful, but not unique. A possible source of information is the US Patent Office. Does anyone know if Audyssey is covered by a issued patent, and if so, its number?
I would appreciate comments, or any available information. A response from Audyssey, or any of their licensees would be most valuable.