"Official" Audyssey thread (FAQ in post #1) - Page 1866

Thanks Markus, tha't good material, but I would have been more interested in your pre-optimization results based on single point measurements. As far as I can remember, nobody has ever done such evaluation prior to running Audyssey, so it would be interesting to see how you are making progress for this subject.

Here is a single point measurement at the primary LP:



Here is a 3-point measurement with the mike 2 ft. on either side of the primary LP. The software I use will calculate an average of the 3 positions that is weighted 40:30:30, with the 40 being the 1st, (primary), measurement position:



The big dip around the crossover frequency is consistent across the "area".
Yes, you've made this comment about my measurements before. I agree with you in principle, but the maximum resolution of the software I was using when I took these measurements was 1/6 octave. There is nothing I can do about it now. Nonetheless, in spite of the lowered resolution, it is abundantly clear that correcting the response at this resolution makes a *huge* difference in the sound. There may be a little more to be had with higher resolution measurements, and the next generation of the software has doubled the resolution to 1/12 octave.

In your opinion what is the ideal resolution for measurement software? There are eight "notes" in an octave, so how finely does one need to measure to delineate those 8 notes? 1/24 octave would be 3 points around each note. Is that "enough"? Can humans "hear" any more resolution than that?

More importantly, how much resolution is needed, (or even possible), on the correction side? I've never seen a parametric EQ with higher resolution than individual frequencies. Audyssey can set thousands of filter taps. Can they set taps at fractions of frequencies? IOW, can they set a tap at 63.25 Hz and another at 63.5 Hz? Is there any benefit in doing so? I don't know the answer, but my sense is that it's not really that important. Of course, I could be wrong.

Your thoughts...?

Craig
Edited by craig john - 8/21/12 at 4:10am

Increased filter resolution will never be a bad thing. It's not a question of need or ideal - technically, you don't *need* to do any EQ to begin with. You can easily hear a difference between 1000hz and 1010hz, so your ear is capable of far better resolution than 1/8 octave. Id take as much measurement resolution as I could get, provided my correction equipment was equally up to the task.

Surely there are diminishing returns, but if you're going to do it, might as well go all the way. There are clearly issues in that room that are brushed under the rug by the smoothed graph.

I have a question regarding the Audyssey MultEQ Movie / Music curves + RE-EQ.

I have read through the Audyssey 101/FAQ/Setup Guide and also A Guide to Audyssey in the blu-ray.com forum. I am confused by some Onkyo specific statements in the blu-ray forum guide. I have a 3.0 system and Onkyo TX-NR709 receiver.

Now when I set up my Audyssey, then for inputs Game (Xbox 360) and TV/CD (Sharp with ARC), using the Receiver > Home > Audio > Audyssey menu, I can select Off/Music/Movie curve as also RE-EQ On/OFF independently of each other. Does this conflict with the above 2 quotes which are from Chris Kyriakakis (Audyssey) himself?

Also, when I switch to NET input, using the same above menu, I can select Audyssey Off/Music/Movie curve but now the RE-EQ option is disabled. Is RE-EQ disabled for NET input?

I have never seen the RE-EQ(THX) option? Where is it? Or rather is it not Active because of my 3.0 system?

Craig's purpose was to use the tools he had at hand to tweak the sub distance(s) to smooth the sub/mains splice well beyond, obviously, what is provided by Audyssey. I *thought* he had measured at three points, but if he didn't it's because he is most concerned about the MLP. Having heard Craig's system many times in its various configurations, I can tell you that his sub distance tweak has improved the sound quality everywhere in his listening area.

So, I doubt that Craig sees any "problems" with his data. I don't, nor would anyone else that sets out to duplicate what he (and Seaton) have done.

Jeff

edit: I see I am late to the party and Craig has already dealt with this. I will leave my post to reinforce the point.

I don't have the ability to measure at higher than 1/12th octave resolution ATM. That is the limitation of my software. I am (personally) perfectly happy with this software and what it can do to help integrate my own subwoofers, as well as the other systems on which I've used it. I don't intend to replace it ATM.
I completely agree that aligning levels "does not automatically result in the best equalizable (combined) response?" I don't even use Audyssey's level-matching process. Level-matching will only be correct for one (arbitrary) LP. For any other LP, it will, by definition, be incorrect. I use gain-matching, (set the gains of all subwoofers exactly the same irrespective of the relative levels at the (arbitrary) LP. I have posted the entire process I went through to integrate my subs here: http://www.avsforum.com/t/759877/seaton-sound-submersive1/4500#post_19446901 There are separate, (low resolution), measurements of each sub individually, the combined response of all 3, and the post Audyssey response.

Since that post, I have changed the setup to two subs on one output and one sub on the other output. This impacted the gain-matching in that the subwoofer trims were different for Sub1 and Sub2. IOW, even though the subs were gain-matched prior to running Audyssey, the input signals were different so their outputs were different. To account for this I "split the difference" and set both trims half-way in between the Audyssey-set trim positions, so all 3 subs received the same level input signal. That gets them back to being perfectly "gain-matched."

And I believe you do something different altogether when setting the levels of your subwoofers.

Time alignment? When I had my subs hooked up with all 3 running off of one subwoofer output, I didn't time align them at all. Once again, time alignment is only valid for one, (arbitrary) LP. That one setting will be incorrect for any other LP. Nonetheless, when I re-set the system with 1 sub on Sub1 and 2 the subs on Sub2, I now had the ability to time align them a little better. The 2 subs on Sub2 are time aligned because they're equidistant to the LP. The 3rd sub is time aligned by the use of the subwoofer Distance setting. Then by adjusting the time alignment of the subs independently, I was able to optimize the system, (combined subs + speakers), a little more. IME, the interaction of all the drivers covering the frequency range in question is what needs to be optimized. The speakers will have some contribution below the crossover and the subs will have some contribution above the crossover. Therefore the range above and below the crossover is the important range. I have optimized my system for the primary LP. It is still very good for other LP, but it's not as optimal as the primary. I don't know that there is much more one can do than that.

I think it is very important to note that time alignment of the subs with each other is not that critical, but time alignment of the subs with the speakers for optimization of the crossover is the critical factor.

Craig
Edited by craig john - 8/21/12 at 8:19am

Thanks Jeff!!! You have heard first-hand the effects of these tweaks, so I appreciate you re-inforcing what I'm saying.

Craig

I pretty much leave "those" posts un-commented on, but the constant drum beat is like water torture and occasionally I succumb. And you can quote me on that.

Jeff

Here's what I don't get: What "more" could I do with higher resolution measurements? I'm using gains and distances and Audyssey MultEQ XT32. The combination is doing a stellar job of flattening the response, at least at the resolution I can see/measure. Even if there are some narrow bandwidth peaks or nulls that I'm not seeing, what could I do about them anyway? I had a parametric EQ for a while, but the current system works so well that I got rid of it. So, what exactly could I improve with higher resolution measurements? All I would "know" is that the anomalies exist... without any way to correct them anyway.

Craig

There is no more that "you" could do. But someone looking for research data might find it helpful.

How likely is it that the Audyssey folks could be persuaded to do something about the crossover problem?

Good question! I will pose it to Chris via email, but it would also be a good question for Ask Audyssey.

Jeff

His system sounds STUPENDOUS. Perhaps the best I've ever heard especially the bottom end. Beyond that, there are still placement limitations he has due to .. well the usual reasons in a multipurpose space. Plus, I doubt that he wants anything raised off the floor, so that dimension is "out." He is a happy camper as would be anyone whose system sounded that good. Cajoling and haranguing won't get him to buy different gear and make him spend more time trying to find another jot and tittle of performance.

As we all kow by now Audyssey has nothing to do with Crossover Management coz its the sole territory of the AVR/AVP makers. The crossover block is always after the Audyssey block, so Audyssey (currently) can not do anything with what it doesn't see.

IMHO, the problem is more complex, so a tight cooperation between Audyssey and the AVR/AVP makers would be needed to rectify the crossover problem.

Let's see what Chris says!

Just thinkn' out loud!

I would imagine it would be somewhat difficult because, to perform the test, the sub and the speakers would need to be "pinged" at the same time, and this would need to be done *after* the measurements and EQ have been finalized. .Audyssey currently has no provision to do this. Audyssey can only ping each speaker and subwoofer by itself, and it doesn't re-check anything post-processing. It would likely take some significant re-programming to allow it to do these things

Craig

Craig, and then what would happen after all the crossover smoothing is done and the dear customer decides to choose another c/o frequency on the fly? Pull out the Audyssey mic again? OMG, getting even more complex.

 

That is plain wrong. It is true for some Onkyo products and not for others. My Onkyo 5509, for example, allows you to choose "music" or 'Movie" curves, which are Onkyo terms for "flat" and "Audyssey curve".

 

That is correct - it should also be correct for *any* THX unit.

The first quote is incorrect because it is outdated. The second quote is correct. It does not conflict with what you say - if you engage Re-EQ *and* the 'Movie" curve, then you have double roll-off.

 

 

Not sure - not tried that. It may be that THX modes are not available in NET mode??

What MultEQ doesn't "see" is a system sweep. My hunch, though, is that there might be a way to analyze the individual sweeps and compute the splice. But that is a guess on my part. and if possible, it might need more computational resources than are available.

And, Audyssey is being somewhat cagey whether it's part of Pro. They seem to imply that they do it with their crossover verbiage.... but without an explicit statement that they do phase alignment in the crossover region, I would tend to think that they don't.

It may just be a question of phrasing, but the crossover is chosen by the user prior to the filter computation in Pro. Different crossovers, different filters.

Do you remember the reference off the top of your head?