Originally Posted by Pete
I'm sure there are differences in hardware...Audyssey, after all, has all but exited the hardware side of things when they stopped offering their outboard boxes (balanced and unbalanced)...but I'm not so sure they would agree with some of the above assertions.
Having spent plenty of time on an outboard Audyssey Pro processor, I can tell you it is a crude device as far as visibility into what it is doing and level of control it provides. This is on top of the poor performance we had when we let it do its thing. Three times in a row, would provide three different corrections. And depending on content you played, it would routinely make things worse. We left if off at the end.
"…parametric equalization methods use a particular type of digital filter called Infinite Impulse Response (IIR) that only attempts to correct the magnitude response in the frequency domain.
That is just wrong or marketing speak. Time and frequency domain are completely connected. If you make a change in one, it will make a change in the other. http://en.wikipedia.org/wiki/Fourier_transform"The Fourier transform is a mathematical operation with many applications in physics and engineering that expresses a mathematical function of time as a function of frequency, known as its frequency spectrum; Fourier's theorem guarantees that this can always be done."
IIR filters can have non-linear phase response but that is a very different thing that saying one corrects in one domain and the other, in another as if they are separate universes.
These filters can cause unwanted effects, such as ringing or smearing, in the time domain particularly as the bands get narrower.
Any filter can have "unwanted effects."
MultEQ uses Finite Impulse Response (FIR) filters for equalization that use several hundred coefficients to achieve much higher resolution in the frequency domain than parametric bands. Furthermore, by their nature, FIR filters simultaneously provide correction in the frequency and time domains.
So do IIR filters.
Here is an example:
We clearly see that correction of the frequency response also resulted in reduced ringing in time domain (the two graphs show what happens with fine and coarse grain filters). It is not possible for the frequency domain response of a system to completely change for the better yet its time response remain the same as they allude.
FIR filters had been considered to require too many computational resources.
That's right. So for a given computational power, their filter performance is worse relative to the resolution and sharpness of an IIR filter. You get better phase response and ease of implementation but you give its effectiveness. There is no free lunch
"there is no way to control the sound after it leaves the speaker. However, we can measure the effects of the reflections that arrive after the direct sound by looking at the time domain response. It has a certain pattern to it that will depend on the time of arrival of those reflections. Based on that pattern and the similarity of patterns across multiple measured locations we can identify the problems caused by reflections. Then, a filter is created to invert those problems as best as possible.
Fancy way of saying it is trying to guess what is a speaker issue, and what is a reflection. And doing it based on raw data rather than building a perceptual model into the measurements and then simply acting on that.
The key is to not think of the filtering in the time domain.
Back to saying 1+1 is not 2
It's not like a graphic equalizer that can only raise or lower the amplitude at certain frequencies. Our filters are in the form of impulse responses that operate on the audio signal through an operation called convolution.
No one uses "graphic equalizers" to correct for room response. The problem with them is lack of resolution, not that they are working in one domain or the other.
... is that they operate in the time-frequency domain as well. That means they can be used effectively to lower the ring down time of room modes in the low frequency range."
And as I just showed from the figure in Dr. Toole's room acoustics book, same thing happens if you correct in frequency domain.
Here is the real issue: how does it sound? Isn't that the bottom line and the question asked in this thread? Audyssey has not provide any double blind studies to show the efficacy of the system. If FIR filters and their implementation is so superior, why not produce such a test? They should easily win if the other systems "smear" things in time domain and such. Right?
Well, someone did run the tests and the results presented at Audio Engineering Society:
Title:The Subjective and Objective Evaluation of Room Correction Products
Authors: Sean E. Olive, John Jackson, Allan Devantier, David Hunt and Sean M. Hess
Harman International, Northridge, CA, 91329, USA
Results were such:
RC6 is Audyssey, the worst performing one in this test, making the subjective quality worse than no correction. RC1 and RC2 are the EQ in JBL Synthesis which uses IIR filters with one optimized for all seats and the other, just one :"The more preferred room corrections (RC1-RC3)
produced smoother curves with more extended output
below 60 Hz. The slope of the curve also seems to be an
important indicator of preference: The more preferred
room corrections had more downward sloped curves,
whereas the least preferred room corrections (RC5 and
RC6,) tended to have flatter slopes, probably related to
the manufacturer’s choice of target curve. Most of the
room corrections were able to detect and fix a room
resonance at 48 Hz visible in the uncorrected
loudspeaker/subwoofer (RC4). However, some room
corrections added a peak at 80 Hz. All of the room
corrections except RC6 were able to correct the wide 2
kHz dip in the sound power response of the loudspeaker"
Bottom line: you can't just use signal processing to solve problems here. A microphone is not a replacement for two ears and a brain as Dr. Toole would say. The analysis needs to have a perceptual model and know what needs fixing. Filters are the chisel, not the craftsman building furniture
BTW, it is sad that the manufacturers in this thread have not provided double blind listening tests of their systems either. I hear them saying go and listen but it is hard for most people to get access to all of this equipment and set them up for proper evaluation as Harman did. We would know a lot more about how well they work if they did provide such testing