"Official" Audyssey thread Part II - Page 31 - AVS Forum | Home Theater Discussions And Reviews
Forum Jump: 
 13492Likes
Reply
 
Thread Tools
post #901 of 3816 Old 08-08-2016, 03:13 PM
AVS Forum Special Member
 
mogorf's Avatar
 
Join Date: May 2009
Location: Budapest, Hungary
Posts: 5,610
Mentioned: 8 Post(s)
Tagged: 0 Thread(s)
Quoted: 1127 Post(s)
Liked: 723
Quote:
Originally Posted by donktard View Post
Wouldn't be too time consuming, just take 10 second sweeps and measure in a predictable pattern so you don't have to write/memorize much.
I think the most time consuming part would be the analysis of those 20-30 curves taken during measurment while trying to sort them out and narrowing the batch to 8 pieces of worthwhile mic positions. This idea came up coz of the hardest part: the predictable pattern. Yet, a visual would surely render a tool that'll make the process easier in the end.

Thanks for your thoughts and feedback. Let's keep talkin'!

Last edited by mogorf; 08-08-2016 at 03:21 PM.
mogorf is offline  
Sponsored Links
Advertisement
 
post #902 of 3816 Old 08-08-2016, 03:39 PM
AVS Forum Special Member
 
mogorf's Avatar
 
Join Date: May 2009
Location: Budapest, Hungary
Posts: 5,610
Mentioned: 8 Post(s)
Tagged: 0 Thread(s)
Quoted: 1127 Post(s)
Liked: 723
Breaking news!

After posting this brave idea to Audyssey Tech Talk on FB here's how Chris K. commented (within an hour):

"Pretty cool idea Feri. I would perhaps paraphrase step 5 to say: "look for any measurements that are dramatically different from the rest (e.g., major bass peak or dip, or diminished high frequencies) and avoid measuring there as it may end up skewing the filters".

mogorf is offline  
post #903 of 3816 Old 08-09-2016, 07:56 AM
AVS Forum Special Member
 
mthomas47's Avatar
 
Join Date: Nov 2007
Posts: 5,788
Mentioned: 93 Post(s)
Tagged: 0 Thread(s)
Quoted: 3399 Post(s)
Liked: 5000
Quote:
Originally Posted by mogorf View Post
Breaking news!

After posting this brave idea to Audyssey Tech Talk on FB here's how Chris K. commented (within an hour):

"Pretty cool idea Feri. I would perhaps paraphrase step 5 to say: "look for any measurements that are dramatically different from the rest (e.g., major bass peak or dip, or diminished high frequencies) and avoid measuring there as it may end up skewing the filters".


Hi Feri,

It is a cool idea, and praise from Chris is the ultimate seal of approval. Congratulations!

I gave this some thought and wanted to put down some rambling impressions. First, from the standpoint of it being somewhat time consuming, I don't know that it would be more time consuming than doing multiple random calibrations in the hope of stumbling on a good one. I like the systematic nature of the procedure, and don't really see any downside to it.

Second, I think that the procedure would be primarily useful in assuring users of optimized frequency response, as opposed to necessarily improved sound quality. Here is what I mean by that. We have all seen people chase measurements virtually ad infinitum. I'm not knocking that--chasing measurements is a legitimate aspect of the audio hobby in itself. But there is not always a direct correlation between improved measurements and improved sound. For instance, I have seen people chase a null somewhere in the listening area for weeks, using re-positioning of multiple subs, bass traps, and endless calibrations, without success. I remember one particular instance where it was a very narrow null at about 70Hz. But there was never any evidence that the OP could actually hear anything amiss. That was partly because of the binaural nature of our hearing, and also because even a small movement of the head could be enough to allow the OP to hear around the null. Our brains are also very adept at filling-in missing information, including frequencies, without us even being aware of it.

So, good measurements should equal good sound, but even slightly better measurements, may not equal audibly better sound. The other difficulty I see would be in comparing a decent randomized calibration with an optimized one. That's easy to do graphically, but hard to do in a listening test. The best way to do it would probably be to save both calibrations to a laptop, and then to go back-and-forth between them. But I believe that it takes several minutes to switch calibrations, so audio memory would not be our friend in that case, and any kind of blind test would be even more difficult to manage.

I trust my ability to hear a bad calibration, and I recently used the example of getting the Audyssey mic too close to a fluffy blanket as an example of that. It gave the upper mid-range a harsh sound that was clearly audible to me, and which was corrected when I did a more diligent calibration. So, I can hear the difference between a really bad calibration and a good one, but can I hear the difference between a good calibration and a somewhat better one? I'm not sure that I can. I think that the difference would have to be at least a little overt in some area of the frequency range, whether it was bass, mid-range, or treble. A little smoother sound could easily escape audible detection, or be easily attributed to expectation bias.

As I said at the beginning, these are sort of rambling thoughts. And they are not intended to diminish the value of the idea, in any way. If a person has measuring capabilities, and a little time to do it, I don't know why he wouldn't want to try the new procedure. Well, I suppose that already having an outstanding frequency response might be a good reason. And for someone just starting out with Audyssey, I can't think of any reason at all not to try it. But I will be very interested, if one of the long-time Audyssey users, such as Alan tries it, to find out how much improvement in frequency response is achievable, and whether the improvement seems audible, as well.

Meanwhile, Feri, keep the ideas coming.

Regards,
Mike
lovingdvd, audiofan1 and mogorf like this.
mthomas47 is online now  
 
post #904 of 3816 Old 08-09-2016, 08:37 AM
AVS Forum Special Member
 
Join Date: Sep 2015
Location: Europe, Croatia
Posts: 1,237
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 720 Post(s)
Liked: 379
Personally, I don't think picking 8 specific measurements is necessary. If nothing, it might prove to be counterproductive depending on your choice (sort of like when you do 1 point measurement 8 times).
8-9 points of relatively tight measurements in a rectangle shape around listening area (+ middle) would be enough for Audyssey to make a decent average, and ONE weird measurement will not affect overall averaging much, but more then one uncharacteristic measurement might do much more damage.
So instead of actually picking very specific points for measuring, we should simply specify zones/areas to avoid. Minimum proximity to back seat would be one choice of zone to avoid.
mogorf likes this.

Revel Ultima Salon 2, Revel M106, Revel C208, Yamaha P5000S, Denon X5200, Panasonic 65" VT50, SVS PC13-Ultra
Kef LS50, Parasound New Classic, Focusrite 2i2
donktard is offline  
post #905 of 3816 Old 08-09-2016, 10:13 AM
Wireless member
 
pepar's Avatar
 
Join Date: Jul 2002
Location: On the beach in Quintana Roo
Posts: 26,287
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 1074 Post(s)
Liked: 551
Quote:
Originally Posted by mogorf View Post
Breaking news!

After posting this brave idea to Audyssey Tech Talk on FB here's how Chris K. commented (within an hour):

"Pretty cool idea Feri. I would perhaps paraphrase step 5 to say: "look for any measurements that are dramatically different from the rest (e.g., major bass peak or dip, or diminished high frequencies) and avoid measuring there as it may end up skewing the filters".

This confirms what most of us discovered with experience; changing measurement positions WITHIN the given pattern criteria affected our results, sometimes dramatically so. Now we have a term (and an explanation) for it: Skewing our filters.

Thanks, Feri, for your diligence!

Jeff

"The future is already here, it's just not evenly distributed." W. Gibson

"I like the future, I'm in it." F. Theater
pepar is online now  
post #906 of 3816 Old 08-09-2016, 10:24 AM
Wireless member
 
pepar's Avatar
 
Join Date: Jul 2002
Location: On the beach in Quintana Roo
Posts: 26,287
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 1074 Post(s)
Liked: 551
Quote:
Originally Posted by donktard View Post
I do have one, regarding sub distance tweak.

1. Reset sub to default state
2. Run Audyssey only because it will set default distances/levels
3. Turn Audyssey EQ Curve off
4. Do sub distance tweak wtih crossover at 80 Hz (predicted best setting so far)
5. Set all speakers to large and LFE mode to sub + mains and also set LPF to highest value. This is so Dirac can measure all speakers+sub full range (otherwise it would measure crossover slope)
6. Run Dirac calibration
7. Set speakers to small and apply crossovers (if necessary, never used Dirac multichannel version so I don't know yet how it handles that)
7. Measure results and do sub tweak again if necessary.

My question is, is step (4) redundant, so I should only do sub distance tweak after full calibration + settings are done?
5, 6, 7, 8 - Dirac 1) tests all speakers at full range regardless of the small/large setting, 2) ignores distance settings and 3) does not give access to the distance settings in the Dirac setup menu.

I am not certain, but I think a Dirac calibration is voided if you change from large to small.

"The future is already here, it's just not evenly distributed." W. Gibson

"I like the future, I'm in it." F. Theater
pepar is online now  
post #907 of 3816 Old 08-09-2016, 10:33 AM
AVS Forum Addicted Member
 
lovingdvd's Avatar
 
Join Date: Jul 2002
Posts: 12,094
Mentioned: 51 Post(s)
Tagged: 0 Thread(s)
Quoted: 2861 Post(s)
Liked: 608
Quote:
Originally Posted by mogorf View Post
Breaking news!

After posting this brave idea to Audyssey Tech Talk on FB here's how Chris K. commented (within an hour):

"Pretty cool idea Feri. I would perhaps paraphrase step 5 to say: "look for any measurements that are dramatically different from the rest (e.g., major bass peak or dip, or diminished high frequencies) and avoid measuring there as it may end up skewing the filters".

Great to hear it. Are you going to update your original post to reflect his advice?

I think the greatest challenge to using this approach in practice is getting the mics positioned the same. For example you could find the 8 spots with REW that measured the most consistently. But when you go to put the Audyssey mic in that spot, even if you are very close, say within an inch, its possible that just that inch difference could yield an entirely different measurement. Granted, this is still far better than picking 8 spots without having any idea whether the area is an anomaly spot or not so it optimizes your chances with your approach.

Now I think the BEST solution is for Chris to implement something like this directly in Audyssey. For instance have it ask you to place the mic in 16 locations, and then have it THROW OUT the most skewed 8 of that 16, and then do averaging with just the 8. So the averaging is done with 8 points like it is now, but with the best 8 points to start with. Who knows perhaps there is already something in Audyssey where it will automatically disregard x of the 8 measures which are too different from the overall average? But if that's the case, it would sure be better to give us 16 points of which it can select the best 8 to average. Want to run that by Chris?
lovingdvd is offline  
post #908 of 3816 Old 08-09-2016, 10:52 AM
AVS Forum Special Member
 
Join Date: Sep 2015
Location: Europe, Croatia
Posts: 1,237
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 720 Post(s)
Liked: 379
Quote:
Originally Posted by pepar View Post
5, 6, 7, 8 - Dirac 1) tests all speakers at full range regardless of the small/large setting, 2) ignores distance settings and 3) does not give access to the distance settings in the Dirac setup menu.

I am not certain, but I think a Dirac calibration is voided if you change from large to small.
How can Dirac avoid my receivers settings?
In my mind it seems exactly the same like measuring with REW, so like, if I try to measure a speaker with crossover on AVR set to 100 Hz, I will get incorrect full range measurement below 100 Hz.
Also, I don't intend to change distance in Dirac but in AVR, though regarding that, Dirac has delay setting for each channel which you can change, which is essentially the same.

Revel Ultima Salon 2, Revel M106, Revel C208, Yamaha P5000S, Denon X5200, Panasonic 65" VT50, SVS PC13-Ultra
Kef LS50, Parasound New Classic, Focusrite 2i2
donktard is offline  
post #909 of 3816 Old 08-09-2016, 10:59 AM
AVS Forum Addicted Member
 
lovingdvd's Avatar
 
Join Date: Jul 2002
Posts: 12,094
Mentioned: 51 Post(s)
Tagged: 0 Thread(s)
Quoted: 2861 Post(s)
Liked: 608
Quote:
Originally Posted by donktard View Post
How can Dirac avoid my receivers settings?
In my mind it seems exactly the same like measuring with REW, so like, if I try to measure a speaker with crossover on AVR set to 100 Hz, I will get incorrect full range measurement below 100 Hz.
Also, I don't intend to change distance in Dirac but in AVR, though regarding that, Dirac has delay setting for each channel which you can change, which is essentially the same.
Is it possible to use Dirac with the 8802a?
lovingdvd is offline  
post #910 of 3816 Old 08-09-2016, 11:04 AM
AVS Forum Special Member
 
Join Date: Sep 2015
Location: Europe, Croatia
Posts: 1,237
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 720 Post(s)
Liked: 379
Quote:
Originally Posted by lovingdvd View Post
Is it possible to use Dirac with the 8802a?
I have DIRAC on my HTPC. Software only. That means that only content played from that HTPC will receive room correction benefits. Which is fine by me since HTPC is my only (important) source of movies/music.

Revel Ultima Salon 2, Revel M106, Revel C208, Yamaha P5000S, Denon X5200, Panasonic 65" VT50, SVS PC13-Ultra
Kef LS50, Parasound New Classic, Focusrite 2i2
donktard is offline  
post #911 of 3816 Old 08-09-2016, 11:26 AM
Wireless member
 
pepar's Avatar
 
Join Date: Jul 2002
Location: On the beach in Quintana Roo
Posts: 26,287
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 1074 Post(s)
Liked: 551
Quote:
Originally Posted by donktard View Post
How can Dirac avoid my receivers settings?
In my mind it seems exactly the same like measuring with REW, so like, if I try to measure a speaker with crossover on AVR set to 100 Hz, I will get incorrect full range measurement below 100 Hz.
Also, I don't intend to change distance in Dirac but in AVR, though regarding that, Dirac has delay setting for each channel which you can change, which is essentially the same.
Quote:
Originally Posted by donktard View Post
I have DIRAC on my HTPC. Software only. That means that only content played from that HTPC will receive room correction benefits. Which is fine by me since HTPC is my only (important) source of movies/music.
OK, I re-read and see you have an off-AVR Dirac solution. So it would be dependent upon the AVR's size/crossover settings. I would guess that it's distance settings are additive to the AVR's.

Am still clueless as to why you'd do an Audyssey calibration for setup details that you can do manually. And even more puzzling is why you'd use spkr settings for the calibration that are totally different from your normal settings.

Jeff

"The future is already here, it's just not evenly distributed." W. Gibson

"I like the future, I'm in it." F. Theater

Last edited by pepar; 08-09-2016 at 11:34 AM.
pepar is online now  
post #912 of 3816 Old 08-09-2016, 11:28 AM
Wireless member
 
pepar's Avatar
 
Join Date: Jul 2002
Location: On the beach in Quintana Roo
Posts: 26,287
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 1074 Post(s)
Liked: 551
Quote:
Originally Posted by lovingdvd View Post
Is it possible to use Dirac with the 8802a?
**miniDSP DDRC-88A Official Thread**8-channel AI/AO Dirac Live in a box

"The future is already here, it's just not evenly distributed." W. Gibson

"I like the future, I'm in it." F. Theater
pepar is online now  
post #913 of 3816 Old 08-09-2016, 11:54 AM
AVS Forum Special Member
 
Join Date: Sep 2015
Location: Europe, Croatia
Posts: 1,237
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 720 Post(s)
Liked: 379
Quote:
Originally Posted by pepar View Post
Am still clueless as to why you'd do an Audyssey calibration for setup details that you can do manually. And even more puzzling is why you'd use spkr settings for the calibration that are totally different from your normal settings.

Jeff
Why do manually what you can do automatically? Sometimes I wish I had a script for every job in life that takes more then 30 seconds.
I am switching speakers to large in order to measure their full range response and then after optimizing filters I'd manually set crossovers to my preference. If you think its better idea to first set crossovers and then measure and apply filters, I'd really like to know reasons why.

Revel Ultima Salon 2, Revel M106, Revel C208, Yamaha P5000S, Denon X5200, Panasonic 65" VT50, SVS PC13-Ultra
Kef LS50, Parasound New Classic, Focusrite 2i2
donktard is offline  
post #914 of 3816 Old 08-09-2016, 12:19 PM
AVS Forum Special Member
 
mthomas47's Avatar
 
Join Date: Nov 2007
Posts: 5,788
Mentioned: 93 Post(s)
Tagged: 0 Thread(s)
Quoted: 3399 Post(s)
Liked: 5000
Quote:
Originally Posted by donktard View Post
Personally, I don't think picking 8 specific measurements is necessary. If nothing, it might prove to be counterproductive depending on your choice (sort of like when you do 1 point measurement 8 times).
8-9 points of relatively tight measurements in a rectangle shape around listening area (+ middle) would be enough for Audyssey to make a decent average, and ONE weird measurement will not affect overall averaging much, but more then one uncharacteristic measurement might do much more damage.
So instead of actually picking very specific points for measuring, we should simply specify zones/areas to avoid. Minimum proximity to back seat would be one choice of zone to avoid.
Quote:
Originally Posted by lovingdvd View Post
Great to hear it. Are you going to update your original post to reflect his advice?

I think the greatest challenge to using this approach in practice is getting the mics positioned the same. For example you could find the 8 spots with REW that measured the most consistently. But when you go to put the Audyssey mic in that spot, even if you are very close, say within an inch, its possible that just that inch difference could yield an entirely different measurement. Granted, this is still far better than picking 8 spots without having any idea whether the area is an anomaly spot or not so it optimizes your chances with your approach.

Now I think the BEST solution is for Chris to implement something like this directly in Audyssey. For instance have it ask you to place the mic in 16 locations, and then have it THROW OUT the most skewed 8 of that 16, and then do averaging with just the 8. So the averaging is done with 8 points like it is now, but with the best 8 points to start with. Who knows perhaps there is already something in Audyssey where it will automatically disregard x of the 8 measures which are too different from the overall average? But if that's the case, it would sure be better to give us 16 points of which it can select the best 8 to average. Want to run that by Chris?
I think that Donktard's got the right idea here, and it seems to me quite similar to what Chris was saying. It's not so much a matter of picking 8 specific mic positions (down to the 1/4") as much as it is trying to avoid anomalous areas, within the overall listening space, that could adversely skew the calibration.

Audyssey actually does use it's own weighted average algorithm to attempt to give less importance to single aberrant measurements. And reports from people who experimented with Pro, suggested that using more than 8 measurements didn't add a lot to the value of the calibration. So, I think that Audyssey got that part right with the 8 measurement protocol. As Jeff said, it's interesting that Feri developed a procedure to assist in doing what many of us have had to do by trial-and-error. That is to find a microphone pattern which is relatively free of anomalous peaks or dips that would tend to skew the calibration. I remember some posters (Austinjerry, for one) who went to amazing lengths to be able to exactly repeat calibrations. I believe that he suspended strings from the ceiling at exact locations and heights, and taped them up out of the way when they weren't in use.

I don't think that I ever repeat calibrations with that kind of accuracy. I know the pattern that I want to use, and the area that I want to EQ. And then I really concentrate on trying to repeat mic position 1, if I can, and I try to keep the mic at ear height, and to be careful about proximity to any hard (or absorbent) surfaces. But whether I come out to each side by 10" or 11" varies a bit, and whether I come forward by about 18" or 20" also varies.

So, I think your point is well taken about the difficulty of exactly matching mic positions, but based on my experience, once you know the area you want to EQ, and the general pattern you want to use, exact precision isn't necessary.

Regards,
Mike
garygarrison likes this.
mthomas47 is online now  
post #915 of 3816 Old 08-09-2016, 12:57 PM
Wireless member
 
pepar's Avatar
 
Join Date: Jul 2002
Location: On the beach in Quintana Roo
Posts: 26,287
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 1074 Post(s)
Liked: 551
Quote:
Originally Posted by donktard View Post
Why do manually what you can do automatically? Sometimes I wish I had a script for every job in life that takes more then 30 seconds.
I am switching speakers to large in order to measure their full range response and then after optimizing filters I'd manually set crossovers to my preference. If you think its better idea to first set crossovers and then measure and apply filters, I'd really like to know reasons why.
Yes, a macro for life!

I am only familiar with the Dirac onboard the Emotiva XMC-1. That's a totally different animal than a PC-based solution.

Jeff

"The future is already here, it's just not evenly distributed." W. Gibson

"I like the future, I'm in it." F. Theater
pepar is online now  
post #916 of 3816 Old 08-09-2016, 01:34 PM
AVS Forum Special Member
 
Join Date: Sep 2015
Location: Europe, Croatia
Posts: 1,237
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 720 Post(s)
Liked: 379
Quote:
Originally Posted by pepar View Post
Yes, a macro for life!

I am only familiar with the Dirac onboard the Emotiva XMC-1. That's a totally different animal than a PC-based solution.

Jeff
True, I will basically be using Dirac + Audyssey at the same time, with Dirac providing EQ curve and Audyssey everything else. Hopefully.

Revel Ultima Salon 2, Revel M106, Revel C208, Yamaha P5000S, Denon X5200, Panasonic 65" VT50, SVS PC13-Ultra
Kef LS50, Parasound New Classic, Focusrite 2i2
donktard is offline  
post #917 of 3816 Old 08-09-2016, 01:51 PM
Wireless member
 
pepar's Avatar
 
Join Date: Jul 2002
Location: On the beach in Quintana Roo
Posts: 26,287
Mentioned: 7 Post(s)
Tagged: 0 Thread(s)
Quoted: 1074 Post(s)
Liked: 551
Quote:
Originally Posted by donktard View Post
True, I will basically be using Dirac + Audyssey at the same time, with Dirac providing EQ curve and Audyssey everything else. Hopefully.
I suppose this is a dumb question, but I have never seen it come up; when you turn Audyssey off, the distance and level settings are still in effect?

"The future is already here, it's just not evenly distributed." W. Gibson

"I like the future, I'm in it." F. Theater
pepar is online now  
post #918 of 3816 Old 08-09-2016, 02:09 PM
AVS Forum Special Member
 
mthomas47's Avatar
 
Join Date: Nov 2007
Posts: 5,788
Mentioned: 93 Post(s)
Tagged: 0 Thread(s)
Quoted: 3399 Post(s)
Liked: 5000
Quote:
Originally Posted by pepar View Post
I suppose this is a dumb question, but I have never seen it come up; when you turn Audyssey off, the distance and level settings are still in effect?

As far as I have ever been able to tell, those settings are unaffected by turning off the EQ, in the same way that the EQ is unaffected by adjusting trim levels or distances. I think that they are entirely different software programs. I have never left Audyssey off for long in my system, but I'm sure that the trim levels and arrival times didn't change when I did.
mthomas47 is online now  
post #919 of 3816 Old 08-09-2016, 02:12 PM
AVS Forum Special Member
 
Join Date: Sep 2015
Location: Europe, Croatia
Posts: 1,237
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 720 Post(s)
Liked: 379
Quote:
Originally Posted by pepar View Post
I suppose this is a dumb question, but I have never seen it come up; when you turn Audyssey off, the distance and level settings are still in effect?
Yes, you are only choosing between EQ modes: Off, Reference, Flat, Bypass L/R, and whatever else I missed. But everything else, distances, levels, crossovers stays intact. Only if you choose Direct or Pure Direct mode you will bypass everything and only your fronts will play.

Revel Ultima Salon 2, Revel M106, Revel C208, Yamaha P5000S, Denon X5200, Panasonic 65" VT50, SVS PC13-Ultra
Kef LS50, Parasound New Classic, Focusrite 2i2
donktard is offline  
post #920 of 3816 Old 08-09-2016, 02:30 PM
AVS Forum Special Member
 
mogorf's Avatar
 
Join Date: May 2009
Location: Budapest, Hungary
Posts: 5,610
Mentioned: 8 Post(s)
Tagged: 0 Thread(s)
Quoted: 1127 Post(s)
Liked: 723
Thank you guys for all your constructive comments and further ideas on this new procedure on the mic placement scheme.

Let me do some more brainstorming for further discussions and to keep the ball running.

We all know that this procedure is not an EQ'ing procedure, just serves the purpose of "mapping" our listening area in order to rule out abnormalities like heavy peaks & dips in the bass region and diminishing highs in order not to feed MultEQ any junk that will skew up the filters.

Now comes the simplification of matters for determining the proper mic spots after #1 at MLP. Well, having a bunch of graphs will all different kinds of wierd shapes may look a bit scary at the beginning, but again, nothing to be afraid of, coz this is just a drill.

The key to success here should lie in the shape of the graph, nothing but the shape. At this stage we need not worry whether we found a narrow dip at 65 Hz or a huge peak at 45 Hz, doesn't really matter. What matters IMHO is how the shape of the curve looks like at each measurment point and to what extend it deviates from flat. Yeah, flat from 10 Hz to 24 kHz, coz now we have a visual of each spot in front of our eyes we didn't have before when we had to do a lot of trial'n'error based calibrations, even with strings hanging from the ceiling for the mic spots (I remember Jerry doing that) and also remember some guys running Audyssey on a weekly basis experimenting endlessly.

And now in order to determine which mic spots to use and which ones to avoid, we can call "fuzzy logic" into the game as follows:

Set up three groups based on the flatness of the curves:

1. Group 1: Slight deviation from flat
2. Group 2: Medium deviation from flat
3. Group 3: Strong deviation from flat

Remember, here we really don't need to care about exact deviations, we do not need to worry about that crazy dip at a certain frecuency. The most important thing is that now we have a group of three different kinds of deviations from flat. We can omit Group 2) and Group3), select 6 points for MultEQ or 8 points for MultEQ XT and MultEQ XT32 from Group1) and run the Audyssey setup routine as usual.

Even though some might already have a calibrated mic (UMIK-1 or the like) used for REW, but I think at this early stage of acoustical mapping of our seating area this can even be done with the little Audyssey mic. We are not calibrating, just looking at relative differences among the curves and for that purpose the Eiffel tower Audyssey mic should be just fine. No surgical precision, no mathematical exactness, and finally no sweat needed here at this initial stage.

Let's talk!

Further to all the above I've been thinking on how to simplify the measurement part of this scheme. Probably instead of doing a myriad of spot measurements around the seating area it might be more practical to do it with REW's RTA feature.

It would allow to simply pull the mic about and watch how the FR changes and take note of those spots where the diviation from flat becomes more and more apparent.

This might not be so troublesome when repeatedly done with the three front speakers (LCR).

Probably I'll have time over the weekend to do some initial testing of this method.

Your thoughts are appreciated and welcome, as always.
mthomas47 likes this.

Last edited by mogorf; 08-11-2016 at 01:00 PM.
mogorf is offline  
post #921 of 3816 Old 08-09-2016, 03:10 PM
Advanced Member
 
garygarrison's Avatar
 
Join Date: Aug 2013
Location: The Milky Way
Posts: 958
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 627 Post(s)
Liked: 752
Quote:
Originally Posted by mthomas47 View Post
Audyssey actually does use it's own weighted average algorithm to attempt to give less importance to single aberrant measurements.
Yes, and Audyssey says this is better than ordinary averaging, and I believe them. As most people know, they say this is a kind of "fuzzy logic" (a good thing). Adding Feri's suggested procedure to this might make the process even better. At least the filters won't be SUBAR (Skewed Up Beyond All Recognition).
mthomas47 and mogorf like this.
garygarrison is online now  
post #922 of 3816 Old 08-09-2016, 03:51 PM
AVS Forum Special Member
 
aaranddeeman's Avatar
 
Join Date: Jan 2005
Location: Lover's State
Posts: 3,823
Mentioned: 9 Post(s)
Tagged: 0 Thread(s)
Quoted: 1703 Post(s)
Liked: 619
Quote:
Originally Posted by mogorf View Post
Thank you guys for all your constructive comments and further ideas on this new procedure on the mic placement scheme.

Let me do some more brainstorming for further discussions and to keep the ball running.

We all know that this procedure is not an EQ'ing procedure, just serves the purpose of "mapping" our listening area in order to rule out abnormalities like heavy peaks & dips in the bass region and diminishing highs in order not to feed MultEQ any junk that will skew up the filters.

Now comes the simplification of matters for determining the proper mic spots after #1 at MLP. Well, having a bunch of graphs will all different kinds of wierd shapes may look a bit scary at the beginning, but again, nothing to be afraid of, coz this is just a drill.

The key to success here should lie in the shape of the graph, nothing but the shape. At this stage we need not worry whether we found a narrow dip at 65 Hz or a huge peak at 45 Hz, doesn't really matter. What matters IMHO is how the shape of the curve looks like at each measurment point and to what extend it deviates from flat. Yeah, flat from 10 Hz to 24 kHz, coz now we have a visual of each spot in front of our eyes we didn't have before when we had to do a lot of trial'n'error based calibrations, even with strings hanging from the ceiling for the mic spots (I remember Jerry doing that) and also remember some guys running Audyssey on a weekly basis experimenting endlessly.

And now in order to determine which mic spots to use and which ones to avoid, we can call "fuzzy logic" into the game as follows:

Set up three groups based on the flatness of the curves:

1. Group 1: Slight deviation from flat
2. Group 2: Medium deviation from flat
3. Group 3: Strong deviation from flat

Remember, here we really don't need to care about exact deviations, we do not need to worry about that crazy dip at a certain frecuency. The most important thing is that now we have a group of three different kinds of deviations from flat. We can omit Group 2) and Group3), select 6 points for MultEQ or 8 points for MultEQ XT and MultEQ XT32 from Group1) and run the Audyssey setup routine as usual.

Even though some might already have a calibrated mic (UMIK-1 or the like) used for REW, but I think at this early stage of acoustical mapping of our seating area this can even be done with the little Audyssey mic. We are not calibrating, just looking at relative differences among the curves and for that purpose the Eiffel tower Audyssey mic should be just fine. No surgical precision, no mathematical exactness, and finally no sweat needed here at this initial stage.

Let's talk!
So, when you take the measurement, it's one curve/speaker/position. So if you are measuring LCR that is 3 curves per position. Is this correct?
Or we just take measurement for just one speaker, like center only.
mogorf likes this.
aaranddeeman is offline  
post #923 of 3816 Old 08-09-2016, 03:57 PM
AVS Forum Special Member
 
mogorf's Avatar
 
Join Date: May 2009
Location: Budapest, Hungary
Posts: 5,610
Mentioned: 8 Post(s)
Tagged: 0 Thread(s)
Quoted: 1127 Post(s)
Liked: 723
Quote:
Originally Posted by aaranddeeman View Post
So, when you take the measurement, it's one curve/speaker/position. So if you are measuring LCR that is 3 curves per position. Is this correct?
Or we just take measurement for just one speaker, like center only.
This still needs further considerations. Thanks for raising the issue. We'll see.
mogorf is offline  
post #924 of 3816 Old 08-09-2016, 04:01 PM
AVS Forum Special Member
 
Alan P's Avatar
 
Join Date: Jan 2003
Location: South Dakota
Posts: 9,987
Mentioned: 66 Post(s)
Tagged: 0 Thread(s)
Quoted: 5136 Post(s)
Liked: 3835
Quote:
Originally Posted by aaranddeeman View Post
So, when you take the measurement, it's one curve/speaker/position. So if you are measuring LCR that is 3 curves per position. Is this correct?
Or we just take measurement for just one speaker, like center only.
Oh my...didn't think of that.

This compounds the "pre-measurements" considerably.
Alan P is offline  
post #925 of 3816 Old 08-09-2016, 04:19 PM
AVS Forum Special Member
 
Join Date: Sep 2015
Location: Europe, Croatia
Posts: 1,237
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 720 Post(s)
Liked: 379
Quote:
Originally Posted by mogorf View Post
And now in order to determine which mic spots to use and which ones to avoid, we can call "fuzzy logic" into the game as follows:

Set up three groups based on the flatness of the curves:

1. Group 1: Slight deviation from flat
2. Group 2: Medium deviation from flat
3. Group 3: Strong deviation from flat
I would actually disagree with some of this you wrote.
Basically, at this point you are trying to pick best spots and avoid bad ones. And while that in practice might prove to work just great, the problem with that is that you really have no real reference point to go back to. Your old measurement tactics? Probably not good enough.
I suggest a different approach (EDIT: similar approach, but different GOAL), but this time REALLY time consuming...and complicated to pull off.
As anyone who has ever picked up REW and measurement mic, many of you probably know that if you fix a microphone firmly at a specific location and take dozen of measurements at that very same spot, you will always get exact same measurement with very tiny variations.
How does that help? Well, it helps because that means you can repeat exactly the same measurement pattern and get predictably same result. As long as you devise a nice obsessive-compulsive method of having mic in exact same spots (same direction, height and everything).

Here are the steps:
1. Make measurements at your 30 or how many you want locations.
2. Pick best spots for Audyssey (the ones with least weird variations) and run calibration
3. Measure results with REW at few specific spots of your choosing
4. Pick 7 previously picked spots and ONE new that you know has some big deviations and re-run calibration
5. Measure results with REW again at your reference spots
6. Pick 6 previously picked good spots and TWO with wild variations, re-run calibration again
7. Measure results with REW again at youru reference spots
...
X. Go as far as you can by adding additional "bad spot" for measurement

Now, when you have done all that, you can compare Audyssey results with 0,1,2,3... weird measurement spots in a mix and therefore determine how much really do bad microphone positions skew Audyssey results, or to be precise, how many bad microphone positions is neglectible for a decent final result?
I'd suggest repeating steps (4) and (6) more then once while each time picking one (or two, respectively) new "bad position" to have a bigger statistical sample for variations with small number of bad spots involved in measurement.

I wouldn't recommend "easy" way to do this...using multiple single point measurements + few bad ones because that might skew results in unpredictable way based on manner in which Audyssey does averaging.

FOR ALL THOSE WHO DON'T LIKE TO READ BUNCH OF S**T: I propose that we determine how many bad microphone spots is enough to significantly affect calibration results.

Revel Ultima Salon 2, Revel M106, Revel C208, Yamaha P5000S, Denon X5200, Panasonic 65" VT50, SVS PC13-Ultra
Kef LS50, Parasound New Classic, Focusrite 2i2
donktard is offline  
post #926 of 3816 Old 08-09-2016, 04:24 PM
AVS Forum Special Member
 
Join Date: Sep 2015
Location: Europe, Croatia
Posts: 1,237
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 720 Post(s)
Liked: 379
Quote:
Originally Posted by aaranddeeman View Post
So, when you take the measurement, it's one curve/speaker/position. So if you are measuring LCR that is 3 curves per position. Is this correct?
Or we just take measurement for just one speaker, like center only.
If you are measuring with intent to find best spots for audyssey calibration you should obviously take all your speakers for a ride.

Revel Ultima Salon 2, Revel M106, Revel C208, Yamaha P5000S, Denon X5200, Panasonic 65" VT50, SVS PC13-Ultra
Kef LS50, Parasound New Classic, Focusrite 2i2
donktard is offline  
post #927 of 3816 Old 08-09-2016, 04:49 PM
AVS Forum Special Member
 
aaranddeeman's Avatar
 
Join Date: Jan 2005
Location: Lover's State
Posts: 3,823
Mentioned: 9 Post(s)
Tagged: 0 Thread(s)
Quoted: 1703 Post(s)
Liked: 619
Quote:
Originally Posted by donktard View Post
I would actually disagree with some of this you wrote.
Basically, at this point you are trying to pick best spots and avoid bad ones. And while that in practice might prove to work just great, the problem with that is that you really have no real reference point to go back to. Your old measurement tactics? Probably not good enough.
I suggest a different approach (EDIT: similar approach, but different GOAL), but this time REALLY time consuming...and complicated to pull off.
As anyone who has ever picked up REW and measurement mic, many of you probably know that if you fix a microphone firmly at a specific location and take dozen of measurements at that very same spot, you will always get exact same measurement with very tiny variations.
How does that help? Well, it helps because that means you can repeat exactly the same measurement pattern and get predictably same result. As long as you devise a nice obsessive-compulsive method of having mic in exact same spots (same direction, height and everything).

Here are the steps:
1. Make measurements at your 30 or how many you want locations.
2. Pick best spots for Audyssey (the ones with least weird variations) and run calibration
3. Measure results with REW at few specific spots of your choosing
4. Pick 7 previously picked spots and ONE new that you know has some big deviations and re-run calibration
5. Measure results with REW again at your reference spots
6. Pick 6 previously picked good spots and TWO with wild variations, re-run calibration again
7. Measure results with REW again at youru reference spots
...
X. Go as far as you can by adding additional "bad spot" for measurement

Now, when you have done all that, you can compare Audyssey results with 0,1,2,3... weird measurement spots in a mix and therefore determine how much really do bad microphone positions skew Audyssey results, or to be precise, how many bad microphone positions is neglectible for a decent final result?
I'd suggest repeating steps (4) and (6) more then once while each time picking one (or two, respectively) new "bad position" to have a bigger statistical sample for variations with small number of bad spots involved in measurement.

I wouldn't recommend "easy" way to do this...using multiple single point measurements + few bad ones because that might skew results in unpredictable way based on manner in which Audyssey does averaging.

FOR ALL THOSE WHO DON'T LIKE TO READ BUNCH OF S**T: I propose that we determine how many bad microphone spots is enough to significantly affect calibration results.
And all this while I was thinking Audyssey should make your life easier. No hassle where your MLP is with respect to ideal, just run that damn Audyssey for your actual (compromised) seating positions (6-8 of them) and let Audyssey work it's magic to immerse you in Audio nirvana..
Oh well.. When do I watch that movie?
pepar, mthomas47 and donktard like this.
aaranddeeman is offline  
post #928 of 3816 Old 08-09-2016, 05:12 PM
AVS Forum Special Member
 
Join Date: Sep 2015
Location: Europe, Croatia
Posts: 1,237
Mentioned: 3 Post(s)
Tagged: 0 Thread(s)
Quoted: 720 Post(s)
Liked: 379
Quote:
Originally Posted by aaranddeeman View Post
And all this while I was thinking Audyssey should make your life easier. No hassle where your MLP is with respect to ideal, just run that damn Audyssey for your actual (compromised) seating positions (6-8 of them) and let Audyssey work it's magic to immerse you in Audio nirvana..
Oh well.. When do I watch that movie?
Nah, honestly, Audyssey does great job as is...this what we talking about is already a nitpicking. Typically, running calibration and REW measurement afterwards for some tweaks gets the job done in very satisfying manner.

Revel Ultima Salon 2, Revel M106, Revel C208, Yamaha P5000S, Denon X5200, Panasonic 65" VT50, SVS PC13-Ultra
Kef LS50, Parasound New Classic, Focusrite 2i2
donktard is offline  
post #929 of 3816 Old 08-09-2016, 05:32 PM
AVS Forum Addicted Member
 
lovingdvd's Avatar
 
Join Date: Jul 2002
Posts: 12,094
Mentioned: 51 Post(s)
Tagged: 0 Thread(s)
Quoted: 2861 Post(s)
Liked: 608
Quote:
Originally Posted by donktard View Post
True, I will basically be using Dirac + Audyssey at the same time, with Dirac providing EQ curve and Audyssey everything else. Hopefully.
Can you elaborate more on your plan to do this and what the benefit is that you anticipate getting compared to just using Dirac or Audyssey along?

Quote:
Originally Posted by donktard View Post
If you are measuring with intent to find best spots for audyssey calibration you should obviously take all your speakers for a ride.
Would it be sufficient to measure L+R+sub at once and look for the spots near the MLP with the least anomalies?
lovingdvd is offline  
post #930 of 3816 Old 08-09-2016, 05:46 PM
AVS Forum Addicted Member
 
lovingdvd's Avatar
 
Join Date: Jul 2002
Posts: 12,094
Mentioned: 51 Post(s)
Tagged: 0 Thread(s)
Quoted: 2861 Post(s)
Liked: 608
Quote:
Originally Posted by Alan P View Post
Sorry, I wasn't on AVS for most of the weekend.

I see that Mike and others have answered most of your questions and you seem to have a calibration you are happy with.

If there's anything that you feel is still unanswered, let me know!
Can you please clarify what you meant in an earlier post about using REW to check the trims set by Audyssey and which you'd trust more? After an 8 point calibration I use REW for my 2.1 setup and the Right channel was 0.5 dB low. I know, I know... But a half dB is a half dB! It's also quite possible that my mic position is not in the same exact spot which is accounting for the difference.

On a related note, after the calibration when I check the SPL with the MV at 0 using internal test tones and REW, both the R/L channels measure about 73.5 dB, so about 1.5 dB lower than the 75 dB I'm expecting. No biggie. But should I just manually raise them so they are at 75 dB? Strangely enough my subwoofer measures about 8 dB below that. I've adjusted the gain accordingly. But didn't expect that.

BTW why is it that folks say not to put much belief/value in the Audyssey EQ graphs when checking the results? It seems like useful information to me. Are they graphs not accurate or something? It seems to provide a good idea of where it is boosting and cutting, so why do folks discredit these charts? Also - it displays a L and R graph for my 2.1 setup, but I don't see any such graph for the subwoofer. Is there a way you are supposed to scroll to that on the next screen or something, or do they not make the subwoofer EQ graph available?

I want to add that the system sounds amazing with the calibrations I tried (except the 1 point cali which I originally did which sounded too bright). So this is just a matter of trying to take the experience from a 10 to an "11". . Its also a matter of trying to learn the ins and outs of Audyssey as practice for my new dedicated theater room which hopefully will be ready around the end of the month.

Thanks!
lovingdvd is offline  
Sponsored Links
Advertisement
 
Reply Receivers, Amps, and Processors

Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off