"Official" Audyssey thread (FAQ in post #1) - Page 61

Noah,

I guess beauty is in the eye of the beholder

There was a big peak centered at 70 Hz and extended out from 60-80 Hz. That peak was reduced in width after the correction. It was not reduced in height.

However, the range between 150 Hz and 250 Hz or so had a big dip in the before plot. In the after plot you can see it coming up to the 0 dB line with some remaining ripple.

The weighting I am referring to has to do with how these plots are generated. They are not a single mic measurement before and after. Rather, they are a combination of multiple measurements. The easy thing to do would be to just average the measurements and show the average plot. But that would be wrong, because it assumes that each measurement contributes equally. MultEQ uses information from each measurement to decide what problems to attack first and then combines the measurements using a method called clustering that is not based on simple averaging. To make things more exciting, all this is done in the time domain and not the frequency domain that you see plotted.

Chris

Yes, Audyssey works in the time domain, but only to correct the FR. I does not reduce long decay times or ringing except as the expected side effect of reducing a peak in the FR. I have verified this with spectral decay aka waterfall plots.

Chris,

Is this 2 Khz dip always present in the Audyssey Curve and/or the Flat Curve?
For instance, if you're using Electrostat speakers, then there is no midrange-to-tweeter crossover!
In such an instance, I would like to be able to remove the extra 2 Khz dip.

Adri.

Thanks, I wasn't aware of that and I suppose it "solves" one issue....if I go with an Audyssey product, I'll get the Pro kit too.

No, not at all, I don't even have a "room" yet....the HT room isn't yet designed and built. Right now I have only 2.1 in the family room, no true AVR/prepro even.

Once the room is built and all the other equipment at hand, I don't see buying one Audyssey-enabled product, and one Lexicon, and one Meridian, and one TacT, and one Anthem, etc., etc. in order to try out different room correction technologies, and then reselling those I like least. I keep hoping that people will post before and after FR plots for competing systems, generated independently of Audyssey/Lexicon/etc, so I can see at least the FR results.

If you are building from scratch and they do a good job the actual changes to the FR would be minimal. The idea is to get it right and only use the EQ's as a last resort. You can't base your decisions on other peoples room curves. The most important issue being you may love the curve and not like the way it sounds.

Rob

I guess that I don't yet get it then. How can a filter . . . filter over time?

All filters filter over time since their action is not instantaneous. Think about how analog filters are constructed with inductors and capacitors, both devices that store energy and release it over time.

Mebbe yes, mebbe no. How are you assessing the success other than with your ears?

To Jeff's (pepar) point: we always point to room treatment first. When done correctly it can help address many problems. When combined with MultEQ it can produce outstanding results. Most of the people that are initially displeased often find that there was something amiss in the measurements or in the bass management settings. Some, however, simply prefer a particular sound signature. That will have to be achieved outside of MultEQ and it is certainly possible to do so.

Chris[/quote]

I have recently installed my new Integra 9.8 and have run Audyssey. I thought the results were quite good but I want to insure that they are as good as I can get without Pro calibration. The area that I question is with my room treatment. My room is 15W x 22L x 9H. I have ASC bass traps in all four corners. These base traps have both an absorptive and reflective side. I have the absorptive side facing the room on all four. I thought I read somewhere, have no idea where, that the traps in the back of the room should be turned to have the reflective side face the room. My question is am I using the base traps properly for best sound, especially with Audyssey? I am using the following speakers: Magnepan 1.6 for L/R Front, Magnepan CC3 for Center and MC1s for surrounds. Also using a 110" screen for projector.

There are spreadsheets, complex spreadsheets, into which you can plug all of your room measurements and surface treatments and even your seats. And the spreadsheet will do its best to model your room. However, nothing can take the place of acoustical measurements, even if those measurements are done with something like freeware on a laptop with a calibrated mic. Anything else is just guessing. IMO.

Chris,
My room is a large room (over 6000 cu ft). It is a family room that opens into the kitchen and the dining and then it is very leaky in the kitchen (open stairway, open floor plan etc). My seating position does not have a wall right behind it. It is at the end of the family room open into the dining. One dimension is about 17 ft (you were close) but again it steps in at the seating position.

With ETF, you can gate based on time. I am not sure I fully understand your question here...

"Does ETF allow you to compensate for delay? The measurement you are taking sees the internal DSP and undergoes delay. This could mess up the deconvolution algorithm in the ETF because it could "clip" the first peak of the impulse response."

but if I understand you right, with the correct gate time, (which is what I choose), the clipping will not happen. Please elaborate on this statement "The measurement you are taking sees the internal DSP and undergoes delay". Where does it undergo delay?

I examined the original FR more carefully and also 3-d time decay charts of what Audyssey does and the FR in room with no filters. What I see is this, there are several peaks one as high as 12dB at 40Hz and others not that bad. I believe this is the room mode you are talking about. Now here is the critical part of this analysis...

With manual tweaking, I attack these peaks (or room modes) very carefully one filter at a time. When I carefully and strategically tame one peak, "it affects the other peaks" so their effects are diminished. I do no more than 6dB cut. A brute force method, would simply target all peaks and tame them with several filters at one shot. I believe this is what Audyssey does. Doesn't it? Tame all peaks at one shot.
I believe this should be an iterative process. Tame one peak, measure FR, observe, analyze, tame another peak, iterate until target achieved.
I understand what you mean by target curve. Audyssey achieves the flattest response, but this is not always desirable. I think you and I know why the flattest is not always the best. Aggressive use of filters has it's own drawbacks. Again this is preliminary diagnosis. If I spend a few more hours (maybe next weekend) I will know more. There were too many charts so I can't post all of them here. I sure would like to know if what Audyssey does is an iterative process or applying all filters at one time. If it is brute force, apply all at one time, why not do an iterative process. I maybe treading NDA domain again . I still have to measure the in-room system FR and impulse response with Audyssey. I believe, the FR will look flat.
Thanks,
-Jai

I don't. Tell me why trying to replicate the studio master is not best?

Chris,

"There was a big peak centered at 70 Hz and extended out from 60-80 Hz. That peak was reduced in width after the correction. It was not reduced in height."

The link still isn't working for me, but I think I remember that now that you mention it.

Certainly narrowing it is an improvement.

Is it overgeneralizing to say that for a given peak, its audibility increases with the area under the curve?

As a check on my grasp or lack thereof of how to interpret the after plots, lets say the afters looked identical to the befores.

Can it then be concluded that Audyssey was, on average, unable to improve the response?

Or can changes to the time domain behavior make it sound better even though the FR looks no better?

In my HT I have a 110" retractable screen for my projector. Before running Audyssey should I retract the screen?

Gary,
This is a highly debated topic and there have been numerous discussions on this and it is as old as the hills and oceans.
In short, if ruler flat response can be achieved without all those EQ filters, like what one can do in an ideal room like a studio, yes then that is the best case. However, in real world, no HT room is perfect. To achieve a ruler flat response, would require aggressive use of filters. Call me a purist, I don't prefer that approach. To my ears, an FR that is not ruler flat with fewer/no filters sounds better than a ruler flat response with a lot of filters. I have tried this before and what I tell you is through experience.
What I have tried is with traditional parametric EQs like the Rane. I don't know the nuances of what Audyssey does internally and I don't know how much they can disclose. I wouldn't blame them if they can't as I totally understand NDA issues.
But to me, these filters sound no different from parametric EQs that can also achieve a flat response.
Just to clarify, I still have a dip of about 6dB narrow band at 38Hz with manual EQ and sound is incredible. I don't sit there thinking Oh, what did I lose with that 6dB dip. I can correct that dip with more filters rather than just one that I use. But when I do that, the magic is gone. Again, I am not talking bloated, boomy bass. I am extremely sensitive to bass issues. So I know what I am talking about and I do believe I have good ears.

Today I have run MultEQ again because I wanted to narrow down the sub level to about 0dB.
So I did a three step measuring and then I rewieved the settings found.

The sub was at -5,5 but with my surprise the Surrounds were set at 70Hz.

After adjustng againg the knob on the sub amp, I did a eight step measure and I got the sub to -1,5 (success!!!) and again surprsingly the surround at 80Hz.

I am surprised by the Surronds because I remember Chris saying that everything below 80 Hz was set as Full Band by the Onkyo 875 because the Surrounds should be able to go lower than the center which was found instead as Full Band (the surronds has bigger midrange than the center) and also because in the many other measurements taken in the past the Surronds were always found Full Band. Specs for surrounds are 92Hz-20kHz ±3dB and specs for the Center are the same but the Surrounds have 1 5" woofer while the Center has 2 4" woofer

The surrounds and the center are both Klipsch, RSX-5 and RCx4 models.
Any thought about?

ciao

Ok then I think you are saying you do not want room correction but adjustment to a personal preference.

Jai, do you have any acoustical treatments in your room? If not, and you can fit them into budget and decor, try some bass traps.

Really? I didn't get that from what he said.

Are you going to use the system with the screen up or down? Measure under those conditions.

That is what I figured. I just wanted to be sure. Thanks.

Well I did if that's ok with you.

"To my ears, an FR that is not ruler flat with fewer/no filters sounds better than a ruler flat response"

"I still have a dip of about 6dB narrow band at 38Hz with manual EQ and sound is incredible."

That's not what I said. Exercise caution when you take statements like these below out of context. Read the whole thing and don't get the wrong take-away.

Well, for one thing, he did not specify the deviation from ruler flat "with fewer/no filters." And speaker FR response specs are typically +/-5dB, so a 6dB dip doesn't seem excessive, especially if the dip is narrow. But then I wasn't reading anything else into his words.

Here's a different answer to jmcomp124's answer.

What you hear in a real room at above midrange frequencies is not a flat response but a slowly falling one and the reason for that is that air actually absorbs high frequency sounds, and it absorbs more as distance from the source and room volume increases.

Studio masters are mixed with the knowledge that the film is going to be played in a cinema, a space with a large enough volumeand therefore enough airto cause some high frequency roll off so what you hear at high frequencies in a cinema is not the way the studio master sounded when it was mixed in the studio. If you took the studio master tape and played that in the cinema, it would not sound the same as it did in the studio. The master, as heard in the mixing studio which probably is the size of quite a few HT rooms, will be brighter at high frequencies than it sounds in the cinema which is not only larger but has an even greater area of absorptive surfaces in the space than you will find in the studio and I would expect to find more absorption in the studio than in most people's HT spaces.

Your room is, I suspect, smaller than a cinema so if you had a flat response things would sound a little brighter at the top end than the film sounds in a cinema. I would not be at all surprised to find that the amount of roll off at the top end in the Audyssey curve is based on differences in room response between spaces the size of a cinema and those the size of most home HT areas and the end result will therefore be something a fair bit closer to what you hear at the cinema than what you would get using the flat response option.

Of course, if you're lucky enough to have a HT room the size of a cinema, the solution is simple: use the flat curve option and you will get the sort of sound that you hear at the cinema.

I believe you got the right take-away. Yes, the dip is very narrow and it s more like 5dB. With movies, I don't notice this. I don't use my subs for critical music listening. I think I would catch it, with music.
The general rule I follow, is placement, room treatment and then room correction. With room correction, I am very cautious. I use filters only when they are absolutely necessary.

jmcomp124,

Try a longer gate time on your measurements. 100 ms. is giving you the equivalent of a data point every 10 Hz -- kinda coarse for the sub. A 1000 ms gate would give you a data point every 1 Hz. In the ETF tutorial (pic below), they use 1300 or 1400 ms.

Also, as Chris suggested, take a look at the impulse response and make sure the first part of it isn't getting clipped off. It's probably okay but check it anyway. The gate time sets when the measurement ends but I'm not sure how ETF sets when it begins -- if it's automatic or if you need to manually set it.

That's the philosophy which I follow as well.

Thanks. That makes better sense to me and seems to make the case for using Audyssey's results in home HT even stronger. I have tried using Audyssey, Front, Flat and Off and Audyssey is the clear winner.