AVS Forum banner
161 - 180 of 904 Posts

·
Premium Member
Joined
·
521 Posts
Discussion Starter · #161 · (Edited)
Will these adjustments still work if using Dynamic EQ?
Yep. I have read in at least two Audyssey support tickets (including the one I started) that adjusting delays will not affect the EQ'ing that is done. So assuming you should be fine.
 
  • Like
Reactions: markb1980

·
Registered
Joined
·
343 Posts
I think it's just sorting out the basic impulse response, thus accounting for DSP delays, and hence usually ending up feet further out than reality, but not going as far as tweaking that to align phase at crossover.
If you have your sub at the same distance as another speaker, or temporarely place it next to another speaker like the front left, then make a distance calibration and look at the subs distance and the left speakers distance. The difference between the values are the electric delay in the sub and needs to be taken out of the equation.

If the left speaker says 4 meters and the sub 5 meters then it's a 1 meter delay in the sub. Then subtract 1meter from that 5 meter and make the 300/343 correction on those remaining 4 meters and when done add the electrical delay of 1 meter to the result. Doing it any other way will screw up the result.
 

·
Premium Member
Joined
·
521 Posts
Discussion Starter · #163 · (Edited)
Got a response from Audyssey on some follow up questions today. Some informative bits in here (if not a little defensive).

Font Material property Screenshot Parallel Terrestrial plant


What I gather from this:

1) So its confirmed that it is indeed a decrease that is needed.
2) He mentions a new 12.5% number here (not the old 14.3% number), which confirms what was earlier discovered (1 - 300/343 = 12.5%).
3) He said "we did the best we could with MultEQ-X" which confirms the suspicion that MultEQ-X is compensating for a bug in the AVR.
4) The built-in and mobile app is not developed by them, so it would not be a fix they could apply.
5) Interestingly he seems confident there is no correction in the older Pro software, which seems at odds with what @KMO reported. And not sure what to make of "those systems are working as expected". Maybe thats a clue some AVR's do not have this issue?
6) Its obvious that I am explaining something to him they already know well. And likely why I am getting the door slammed. But I do appreciate him telling me what he did. Sounds like they are struggling to deal with something that is not in their control.
 

·
Registered
Joined
·
7,648 Posts
So I physically measured the distances with a measuring tape, and they were exactly what Audyssey had determined (except my centre channel - that was 6 inches out), which made the adjustments for the other speakers I made last night pretty much unchanged.
 

·
Premium Member
Joined
·
521 Posts
Discussion Starter · #167 ·
So I physically measured the distances with a measuring tape, and they were exactly what Audyssey had determined (except my centre channel - that was 6 inches out), which made the adjustments for the other speakers I made last night pretty much unchanged.
Sorry I am not following this. You mentioned previously (last night?) you had made the adjustments to your distances. How does this new message apply to that? Seems like I am missing some context. If you made the adjustments, your AVR values would be smaller than your measuring tape measurements.
 

·
Registered
Joined
·
12,088 Posts
Fixed. Until the next "update." :)
Michael
I guess that's the update I was expecting. ;)
So the take home lesson is to run Audyssey and, if you're not using -X, reduce the distances of all the speakers by 12.5%. But -X already does this, so you don't have to. Right?
Michael
 
  • Like
Reactions: KMO

·
Premium Member
Joined
·
11,172 Posts
Jeff at Audyssey alluded to this in December 2021:

Starts at 1:16:46

 

·
Registered
Joined
·
1,283 Posts
For HT, my understanding is preferred speaker to use for the sub tweak is the CC (at least that's how I've always done it!), for 2.1 music, then my understanding is you'd use the L+R together.
My system is kind of optimised for multichannel music, although it's not the most common thing I play.

Mixers of real multichannel discrete sources - both films and music - tend to actually put most of the bass in L+R, in my experience.

Dolby Pro Logic II upmix is the main thing that ends up putting lots of bass into C. Obviously human multichannel mixers can do it, but they tend not to.

So you have to try to look at all of L and C and R.

Surrounds don't get much in most modern mixes, but old quadraphonic recordings can certainly drive them hard. Not uncommon to have stuff like the drums being entirely in the rear, like in various tracks from The Doors. But not enough of that to be worth the extra complication of worrying about for time alignment.

Oh, one other source there's a lot of - gaming. Games with 7.1 sound drive all channels, probably harder than most normal mixes. The mixing engines tend to have their own built-in crossover to redirect quite a lot of the bass to the LFE channel though, so not sure what proportion of bass energy goes to the 7 speakers. But whatever there is would go to all of them, at least for world audio. Background music tends to be just L+R, with only the occasional real multichannel or Penteo upmix.
 

·
Registered
Joined
·
1,283 Posts
If you have your sub at the same distance as another speaker, or temporarely place it next to another speaker like the front left, then make a distance calibration and look at the subs distance and the left speakers distance. The difference between the values are the electric delay in the sub and needs to be taken out of the equation.

If the left speaker says 4 meters and the sub 5 meters then it's a 1 meter delay in the sub. Then subtract 1meter from that 5 meter and make the 300/343 correction on those remaining 4 meters and when done add the electrical delay of 1 meter to the result. Doing it any other way will screw up the result.
That's incorrect. The required delay is a combination of DSP delay and actual physical distance.

A non-MultEQ-X Audyssey calibration will have measured the combined delay sum, and converted into an equivalent distance using 343m/s.

We need to get the correct compensating total delay from the AVR, which we only get by using (combined delay sum * 300m/s) as the distance.

The fact that the electrical delay is virtual, and not due to sound travelling, doesn't matter. Indeed, the real speed of sound in the room doesn't matter either. Nothing matters, apart from the total delay we need.

We just need to communicate that delay to the AVR via a "distance", so we always need to use a delay->distance conversion that matches the AVR's distance->delay conversion. The entire delay has to be multiplied by 300m/s to generate a distance. If we're showing a distance to the user, based on 343m/s (potentially including virtual distance for a DSP delay), then the entirety of that distance has to be multiplied by 300/343.
 

·
Registered
Joined
·
1,283 Posts
Interestingly he seems confident there is no correction in the older Pro software, which seems at odds with what @KMO reported. And not sure what to make of "those systems are working as expected". Maybe thats a clue some AVR's do not have this issue?
I've gradually been coming around to the view that the MultEQ Pro is not deliberately fixing this. (But I liked the potential story of it being Lost Knowledge).

The Pro's reported distances are lower than physical, but not as low as being the 12.5% correction. It could just be an error that happens to be in the right direction, rather than being deliberate. Or maybe it is a fudge, but they didn't get the factor right? But the support rep reckons there's no deliberate adjustment, and the factor I calculate is quite variable. Could just be poor latency handling.

My first attempts at testing were to blindly try full 14.3% changes in both directions, and both made it worse. I think that can be reasoned out by the original measurement being around -8% from reality which is close to the -12.5% you actually need. Closer to it than to reality, at least. Moving in either direction to -22% or +6% ended up further away, making it worse.

For -14.3% to improve things, you have to be starting from close to accurate distances. Particularly with 14.3% being a bit too big of a change.

And not sure what to make of "those systems are working as expected". Maybe thats a clue some AVR's do not have this issue?
No, the older systems definitely do have the issue - I did a bunch of tests on the AVR-4308 to confirm that each 1m of distance in the config caused a ~3.3ms delay change, so it's using 300m/s or very close.

I don't see a reason to believe he has knowledge of older AVRs' behaviour. He's just attempting to logically reason that "if the Pro gets the alignment right, it must be because the old AVR is using 343m/s." And maybe he's also thinking "if older AVRs had this issue, wouldn't it have been spotted already?".

The Pro isn't getting it "right", just better than the current Audyssey - which is why I was not impressed by your original graphs. But not as good as my tape measure and 300/343. Hard to say what's going on. It's not precise, so maybe it's just erring in the right direction by accident.

The current Audyssey seems to be precise, which has made the Denon speed of sound more clearly visible, which then means we can clearly determine the adjustment to make the delay accurate.
 

·
Registered
Joined
·
7,648 Posts
Sorry I am not following this. You mentioned previously (last night?) you had made the adjustments to your distances. How does this new message apply to that? Seems like I am missing some context. If you made the adjustments, your AVR values would be smaller than your measuring tape measurements.
I reset, then physically measured to compare what the actual distances were vs. what Audyssey determined from the calibration. I then did the math to reduce the distances based on the tape measurements, so the numbers are now smaller. It just so happened that the Audyssey settings matched the tape measurements for every speaker aside from the centre channel, so most of the reductions were the same as what I had done the night prior.

I think my point was that sometimes Audyssey actually gets things right.
 
  • Like
Reactions: Jeremy Anderson

·
Registered
Joined
·
343 Posts
6) Its obvious that I am explaining something to him they already know well. And likely why I am getting the door slammed. But I do appreciate him telling me what he did. Sounds like they are struggling to deal with something that is not in their control.
Denon buys the Audessey application to be used in their AVR. So Audessey are responsible for measuring the time delays and after calculations applying those to each speaker to make the sound from all speakers to arrive at the mic position at the same time. If they only had to do that it would be no problem.

But Denon have their menu application in the AVR and for some reason they want to present a distance value to the user, and more strangely allows a user to change the calibrated values with their own distance settings. It seems that Denon takes the time values from calibration and converts to distance using 300m/s and present that in the menu, as a too long distance. But as they also want a user to enter his own values they must use the distance value and convert to time for Audessey to use for the time correction for the speakers. If Denon also used 300m/s for that calculation it would be no problem, if a user didn't change the distance values.

But to make sense, Denon must have asked Audessey to use the distance values from the Denon menu and convert them to time by themselves. Audessey use 343m/s and the error then occurs.

When I apply the 300/343 formula to the calibrated values I get the exact tape measure distance to my speakers, so Audessey are using the correct distance to time conversion formula. It is Denon that made a mistake and used 300m/s when converting from time to distance, because it's hard to image Denon using 300m/s when converting to distance but 343m/s when converting back to time and then give that time to the Audessey application.

Asian people have a really hard time admitting to any mistakes they do so there are probably some sort of feud going on whos fault it is and who should correct for the fault. Audessey being an US company have no problem to admit to mistakes or do correction by making one error to correct for another error, but are not the correct way to fix that problem as it could introduce other problems when using other external applications that set distance values into Denons menu system.
 

·
Registered
Joined
·
1,283 Posts
Denon buys the Audessey application to be used in their AVR. So Audessey are responsible for measuring the time delays and after calculations applying those to each speaker to make the sound from all speakers to arrive at the mic position at the same time. If they only had to do that it would be no problem.

But Denon have their menu application in the AVR and for some reason they want to present a distance value to the user, and more strangely allows a user to change the calibrated values with their own distance settings.
Strangely? Not everyone is going to use the built-in calibration, and people may well want to adjust the calibration, eg to get sub phase alignment perfect.

It seems that Denon takes the time values from calibration and converts to distance using 300m/s and present that in the menu, as a too long distance.
Again, no, you're misunderstanding how it works. I'll try to explain again. It's much simpler than whatever's going on in your imagination.

Audyssey are taking the time values measured during calibration and converting to distance. The output from the Audyssey calibration is a distance measurement, because that's what the Denon (and almost all other AVRs?) uses as a configuration parameter.

Denon has always used a speaker distance setting, before auto-calibration was a thing. (Because before auto-calibration, we all used tape measures!)

The error is that Denon converts from distance to delay using the incorrect (unless you listen at -49°C) 300m/s. So whoever states the distance (a manual user or Audyssey) has to give short distance to get a correct compensating delay.

Denon is not anywhere converting from delay to distance, as far as I'm aware.

Asian people have a really hard time admitting to any mistakes they do so there are probably some sort of feud going on whos fault it is and who should correct for the fault.
Maybe, but it's really just a more basic PR thing. When you're working with another company, and there are commercial considerations involved, it's not the done thing to have your customer-facing people saying, "oh yes this was the other company's fault, and we worked around it". Got to soft-soap it. It's always "we worked together to find a solution" in public :whistle: , no matter how sweary the engineers in-house are :mad:. I know this, because I'm one of those engineers they sometimes have to keep away from the public. :censored:

And besides, Audyssey probably don't want it fixed now, because that would just make their life harder. A constant correction for 300 m/s for all Denons is easy - having to track which models use which speed of sound constant is a huge potential source of future errors.
 

·
Registered
Joined
·
343 Posts
Again, no, you're misunderstanding how it works. I'll try to explain again. It's much simpler than whatever's going on in your imagination.....Audyssey are taking the time values measured during calibration and converting to distance. The output from the Audyssey calibration is a distance measurement, because that's what the Denon (and almost all other AVRs?) uses as a configuration parameter.
Then my AVC X3700H with the latest firmware operates in a different way. After Audessey calibration with the internal application the distances displayed are too long when comparing to tape measurement. If I apply 300/343 to the distance values it matches exactly to the centimeter to my center speaker and two of the surrounds, the ones I checked. So the displayed distance value in my 3700's menu, after correction, matches exactly to the actual distance to my speakers.
I just checked again. The measurements in the Denon setup are spot on, when measuring from the speakers to the 1st Audyssey mic position.
So it seems that Denon has made a change in my firmware, probably trying to correct for the error but went the wrong direction and made the distance longer instead of shorter.

So in my case I would need to apply 300/343 to the calibrated Audessey distance to make it match the reality, as if I would have made a tape measurement, but then I need to apply 300/343 once again and enter as the distance so it will display a shorter distance than a tape measure? So any external 3:rd party application that has a distance value to be entered into Denon needs also to use 300/343 before entering it to be correct and will show up shorter in Denons menu? Or am I misunderstanding again, educate me.
 

·
Registered
Joined
·
36 Posts
Isn't this coming to the point where D&M should be going public and announcing what the situation is? If it is a mistake/error across multiple products, then they need to provide a list of the products with the error the method to correct it. As said, it could be done by firmware release but, even if it is, there will be products pre-ethernet connection/update whose owners will still need to know how to correct.
 

·
Registered
Joined
·
1,283 Posts
So it seems that Denon has made a change in my firmware, probably trying to correct for the error but went the wrong direction and made the distance longer instead of shorter.
Oh dear. Sounds like you're certainly seeing something different. Seems my fear about an attempted fix making things more confusing might be true.

I think you'd need to actually make some REW measurements to recheck what it's doing.

My initial guess is that they've chosen to rescale the Audyssey-reported distance before applying it, to compensate for the 300m/s, but they did it in the wrong direction?

(Denon might be doing this because the Audyssey code in the device is locked in, and rather than get a patch from Audyssey, they chose to apply an external correction - the opposite to what Audyssey are doing with the MultEQ-X, where they treat the AVR as locked-in. And they didn't want to change the 300m/s, because that would immediately change everyone's existing calibration for whatever particular distances they've set).

So in my case I would need to apply 300/343 to the calibrated Audessey distance to make it match the reality, as if I would have made a tape measurement, but then I need to apply 300/343 once again and enter as the distance so it will display a shorter distance than a tape measure?
Possibly, yes.

But it's also possible that they also adjusted the speed of sound to be 343m/s too. Maybe they've made two changes that cancel out?

You'd need to recheck the actual speed of sound by trying distances 1m apart, and seeing if 1m causes 3.3ms difference (300m/s) or 2.9ms difference (343m/s).
 

·
Registered
Joined
·
158 Posts
They are asking the question on the MultEQ editor app thread if these delays are an issue with the app also? I have said as far as I have read on here it does effect the app also but just checking I have not got that wrong?
 

·
Premium Member
Joined
·
521 Posts
Discussion Starter · #180 · (Edited)
They are asking the question on the MultEQ editor app thread if these delays are an issue with the app also? I have said as far as I have read on here it does effect the app also but just checking I have not got that wrong?
Yep that is correct. Only the MultEQ-X app seems to compensate for the error. All other apps seem to need the adjustment.

But I will add the caveat that no one should be making changes and walking away. You should really do some validation either with listening tests, and/or measurements with a tool like REW. There are still some questions about what AVR's this might apply to, so leery of steering people down the wrong path.
 
161 - 180 of 904 Posts
Top