Using dual 31band EQ's for LCR to Give Audyssey a Head Start? - Page 2

Quote:

Originally Posted by mogorf 

Quote:

Originally Posted by kbarnes701 

I'm not so sure, Feri. I have managed a very respectable FR using Audyssey Pro and then making manual adjustments based on my OmniMic measurements, such as using the Target Curve editor in Pro, and varying the sub delays in the AVP in order to improve the flatness around the XO frequency, but I am thinking that the ability to make small further adjustments, after running Audyssey, eg by using parametric external EQ, could be a good thing. I can't see how it can harm the Audyssey calibration to boost some frequencies a little in order to get a flatter, or preferable, curve.

In fact, I am already doing this to some extent. The amplifier in the Submersive subs has two different DSP modes - one of them (Pgm2) introduces a 3-4dB~ lift starting at around 40Hz. By running Audyssey with the subs in Pgm1 and then, after running Audyssey, changing to Pgm2, there is a subtle but important flattening of the curve at the bottom end. Audibly it is significantly better.

This graph shows the effect of using Pgm2 in this way:





 

Keith, you also know the general advice of taking out every possible filter (and any additional circuitry) out of the way of Audyssey prior to eq'ing a subwoofer. Crossover set to max., phase knob set to 0, best if sub amp has an LFE input bypassing this kinda stuff, and all that jazz. Now you want to introduce another set of filters into the chain in form of a PEQ. Care to reconsider?

Feri, can you not see the improvement in the bass curve in my graph?  Why on earth would I want to reconsider?

Quote:

Originally Posted by mogorf 

Quote:

Originally Posted by kbarnes701 

Unfair. Feri is Hungarian and his command of English is better than many of the native-English speakers who frequent these forums.

I'm being flattered here Keith, ain't I?

No - it is true, based on some of the posts I read every day.

Quote:

Originally Posted by mogorf 

Quote:

Originally Posted by Gooddoc 

That's advice for folks without measuring gear and it doesn't apply to more complex systems.
Audyssey is not the only tool in the shed for those with the knowledge and measuring gear.

IMHO, it's not a question of having measuring gear or not, but a question of adding additional circuity into the chain or not. My position is that it is not advisable to clutter the system with circuitry that may add unpredictable results like phase distortion that may result in unwanted lessening of audio-fidelity at frequencies affected when it comes to odds adding up. On the contrary, feel free to use a PEQ, but make sure to turn Audyssey off. Dual systems are always harder to handle (and troubleshoot) than single systems. Agree?

Not only does my system measure better using the Pgm2 tweak I mention above, but it also sounds better. What else is there?

Quote:

Originally Posted by beastaudio 

Quote:

Originally Posted by kbarnes701 


I am a huge Audyssey fan too (and my sig is even more revealing than Feri's ) but I agree with you. Audyssey is just a tool and like any tool, in the right hands it can be made to perform better. Audyssey is great for the average guy who just wants to run a few chirps and let the software get him a reasonably good result, right out of the box. In its XT32 form, it is very, very good. But not perfect.


Yes, this is exactly what I am thinking too. A couple of additional filters and I'd be flatter as well. Got to be a worthwhile objective surely?

Absolutely!!! Last night I did some measuring, moving, and distance/XO adjustments, just to see what I could come up with over an hour or so of tweaking. I can post graphs this evening, but essentially, with just a couple filters (XO region especially) I was able to get the best response I have ever seen between all my speakers. I meant to post some graphs, but I got sucked in to just sitting and listening for a while and was more than pleased with what I came up with. The biggest thing I noticed though, was the Audyssey sweep I did last night was DRAMATICALLY different from the one I did previously (at least to my ears, I didn't measure the last response). Bottom line, it kinda goes to prove that your results, all things being equal, can be quite different between Audyssey testings, and for that, I think additional eQ is certainly beneficial to keep your naked response closer to the final objective to begin with.
 

Yes, I think it is wrong for anyone to believe that Audyssey can never be improved upon. In fact I know it is wrong because I (and many others) have improved on Audyssey's results. For example, Audyssey rarely gets the sub distances absolutely correct and by tweaking them post-calibration, almost everyone who has tried this has reported an improvement in the FR at the vital XO region. Of course you do need measuring equipment and the knowledge of how to use it, but these days very many serious hobbyists have REW or an OmniMic and are able to make and interpret the various measurements and graphs. 

Audyssey Pro even provides a built-in tool for changing the results afte calibration by way of the Curve Editor - and again it requires independent measuring gear to be able to use it. 

Audyssey is terrific for those who want to simply run the calibration and then sit back and enjoy their system. They will almost certainly have made very substantial improvements compared with an un-EQd system. But, like everything, it is not perfect and those with the desire, knowledge, skill and patience can effect subtle but significant improvements by doing some careful after-calibration tweaking of the sort we have been discussing. 

There are also the issues with deep, room-induced nulls, which electronic EQ can do nothing of significance about and after running Audyssey, some post-cal measuring will reveal the nulls and then one can contemplate means of eliminating or reducing them, with traps or multiple subs or room rearrangement etc. It is not fair to criticise Audyssey over  something it is scientifically impossible for it to fix, and I am not doing that, but this example does show that Audyssey can simply be one tool in the chest when it comes to seeking the 'perfect response'.

Quote:

Originally Posted by Theresa 

While not perfect it is amazing to me how well Audyssey does given imperfect rooms and speakers.

Absolutely. 

Quote:

Originally Posted by beastaudio 

Came home for lunch, here is a little before and after graph...


Before


After


Keep in mind the graph constraints are much tighter in the second graph. Not too bad at all, but I can still see a few spots that additional EQing would be nice

What smoothing are you using?

Nearest I have to your measured range is:

This is with 1/12th smoothing and I have since improved on the flatness of the curve, but can't find the darn graph (I think it is still on my laptop).

Quote:

Originally Posted by beastaudio 

Quote:

Originally Posted by kbarnes701 

What smoothing are you using?

Nearest I have to your measured range is:



This is with 1/12th smoothing and I have since improved on the flatness of the curve, but can't find the darn graph (I think it is still on my laptop).

1/6th smoothing, but 1/12th was quite similar as far as the +/- range goes. Keep in mind that "after" graph is still pre additional eQ, and that my dB range is quite constricted. Even though it looks kinda ugly, I am still +/-3db across the entire range. I think I could get it even better messing around with that one little area from 400hz to 1.2kHz. I would then reset my low-end house curve, which it appears you have as well

1/6th smoothing makes for a nice looking graph but it doesn’t tell us enough about what is really going on IMO.  Generally the most useful smoothing IMO is 1/24th and that is what I (and others) usually use for any graphs I 'publish' here. I also stick to a 5dB range for the graphs as most people seem to use that which makes it easier to compare. If your graph was using the same smoothing and same dB range, I bet it wouldn't look all that different to mine.

Here's one of mine with 1/6th smoothing (probably made on a different occasion to the 1/2th graph shown before so they won't compare precisely):

I am striving for +/- 3dB across the spectrum (other than the low end where I run the subs a little hot as you can see. I have read that a plus or minus 3dB is probably inaudible, hence my striving to get there. To get it to where it is has taken some effort though with Audyssey Pro, room treatments, OmniMic and additional DSP in my Submersives.

Quote:

Originally Posted by beastaudio 

1/6th smoothing is about the resolution that the human ear can hear as well. I will work on getting some additional 1/12th graphs for you as well. What I was trying to say is the 1/12th graph I was looking at in real-time was within the same range as the 1/6th, so I opted for the 1/6th. As you can see, I care more about having a graph that is easily readable, as opposed to just looking pretty, but I will get it back to 5dB increments for the sake of good comparison. You have a great response btw, but with the subM's, wouldnt you want to expand out the bottom end of the graph to measure down to at least 10hz?

Yes, a graph with 1/24th and 5dB increments would be great - that is what most seem to use.  Yes, you are right, I did do some graphs with the bottom end expanded - I made the 20Hz ones for ease of comparison with my old subs, which had no significant output below 20Hz. Here's one that shows the bass frequencies only. This is with a 100Hz XO to the Submersives, 1/24th smoothing.

Here's the same response with 1/6th smoothing:

Here I am pretty much on the money for +/- 3dB.

Quote:

Originally Posted by beastaudio 

No Kidding! That is wonderful. I will have some time Saturday where I might do a little messing around downstairs, and I am also hoping the EQ comes by then as well. Next two nights I have Willie Nelson and then Killswitch Engage (talk about broad genre selection) so I won't have a whole lot of time (or sobriety) to measure properly.

Sounds like fun!  Enjoy. I'll look forward to seeing your graphs after the EQ is in...

Quote:

Originally Posted by gurkey 

@beastaudio
A note to your signature:
European models mostly do accept banana plugs, but you have to remove the small insert in the middle of the speaker connectors beforehand which are been needed because of European regulations.

And an easy way to remove the plug is to partially screw in a small self-tapping screw and then pull out the plug using pliers. 

I used to work in the fashion photography business and a lot of the European models accepted all manner of plugs   (Is this what Beast meant?? :))

Quote:

Originally Posted by Mark Seaton 

Quote:

Originally Posted by beastaudio 


1/6th smoothing is about the resolution that the human ear can hear as well. I will work on getting some additional 1/12th graphs for you as well. What I was trying to say is the 1/12th graph I was looking at in real-time was within the same range as the 1/6th, so I opted for the 1/6th. As you can see, I care more about having a graph that is easily readable, as opposed to just looking pretty, but I will get it back to 5dB increments for the sake of good comparison. You have a great response btw, but with the subM's, wouldnt you want to expand out the bottom end of the graph to measure down to at least 10hz?

Hi guys,

Quick note on measurement resolution. 1/12th-1/24th octave resolution is of interest below 100-300Hz, where there are much narrower band effects we are concerned with or where subwoofer and speaker location changes can make significant changes to very narrow/high Q issues. When looking at the full range frequency response, 1/6th octave does have a good relation to what we discern with our ears, but for purposes of tailoring the balance/voicing of a system, it can also be very useful to look at 1/3rd octave with a glance at 1/12th octave to confirm what is causing the shape of the 1/3rd octave curve. EQ in the upper ranges should always be done with care and lots of listening and measuring.

While the programs will give you a high resolution plot at high frequencies, often it doesn't mean much unless you have sufficient windowing flexibility to compare direct sound (shorter time window) and long term average (ie RTA). Unless it's clear there is a problem in the speaker itself you are taming, better results are often had with more subtle and broad EQ of the response with wide PEQs or shelf filters. Figuring out how to apply such adjustments can be difficult when only looking at high resolution plots with lots of hair and sharp changes. Heavier smoothing helps visualize trends and lets us look at the overall spectral balance we might be hearing.

Thanks Mark. Very illuminating - esp the comments wrt to 1/3 smoothing.