Audyssey's Next Generation of Room Correction (MultEQ-X)

[chych7]

Well that sucks. I was hoping for a simple and easily correctable user error, rather than a defective room correction system. And it is an arguably disqualifying defect: you said zig (lower the HF) and Audyssey decided to zag (boost the HF). That's in essence, "eff the target curve you assign me, I'll do whatever the hell I want." Have you reported this bug to Audyssey? I would start there. It could be as simple as someone baking the wrong cal file into the program.


IF you have interest (regardless of the defect, you seem to be satisfied with the sound you’re getting, so I’d understand a lack of interest) one quickish way to see where Audyssey is defective is a rough microphone comparison: same speaker, swap mics keeping the capsule at same height, distance, and orientation. I don't think the Audyssey uses a weird pinout, so you should be able to plug it right into any computer or USB sound card mic in that accepts a standard 1/8" TS plug. Or just put them side by side if you have a stand that can do that.

Unfortunately I do not have a computer with a microphone input, or an appropriate USB interface, so I can't test this. In any case, it's pretty simple for me to "fix" the high frequency mismatch in MultEQ-X, either by adding another filter or cutting off the room correction at 4 kHz, so I don't personally feel it's worth spending more effort on it. I did always find it strange Audyssey boosted the highs; happened on every single calibration I've run.

 

[Reverend Slim]

FWIW all three Audyssey mics that have passed through me in the last decade (two review units, one personal purchase AVP) have been just fine. However, I've also seen measurements from others suggesting more variance than in my sample of 3.

I have 4 Audyssey mics here and only one that was obviously off, but we're talking by a HUGE margin. It was the mic that came with my Denon 4500. The post-cal result using it was all highs and no bass, and made my system sound like an old AM radio. Fortunately, I still had the mic from my 5200 and it worked fine. Denon ended up sending me a replacement. Barring that one bad mic, the other 3 produce about the same results every time, though I did mark one of them as my preferred mic because its final levels better matched my SPL meter with post-cal checks. Same EQ results, but with that mic, the trims consistently come out 1dB higher than when I use the other two.

 

[Ata]

I definitely don't know more or better than most people in this thread, but the discussion begs the question...

When comparing the different DRC technologies, don't we have to make measurements that reflect what DRC is trying to correct for? I mean, most DRCs do pre-filtering using FDW and other techniques, thus doing a simple REW swipe or MMM measurement, in one or more locations, will not be representative of what the DRC has done in terms of correcting the speaker and room response?

Of course, for most/all of these technologies the algorithms are proprietary, so we can't know what pre-filtering is being used. But, at high enough frequencies and with a small enough window to get rid of reflections, the graphs should be a lot smoother than what is seen above?

 

[peng]

I'm assuming nothing. I'm reading.



You two may think that, but the actual text on the screen says otherwise. I note your hand-waving is not accompanied by any actual explanation of said alleged differences in scope. So, here's your opportunity to do so: please explain to the group the relevant and material distinctions you see between the following statements:

Group 1:
C (quoted, emph. added):
"Often people want to compare their results to the Audyssey predicted response. THAT is when you have to use the exact same measurement points. People usually don't, then complain that Audyssey is "making things up". Hence, the disclaimer on the correct procedure."
C (quoted, emph. added) To clarify, this is "responding to the checking of the "after" graph that Audyssey displays in the software."
P (quoted, emph. added)
"he suggested using the same 8 points ref. post#413)...when using REW to find out how well each performs in terms of actual results (averaged) versus the corresponding target curves."

I have NOT said any of those you quoted in Group 1 that you are implying, and if not please clarify. I have only suggested that you were referring to something different than the OP's and never commented on what I think is a better way. I can't believe you are making such a big fuss and now seem to be resorting to misquoting..

You may be reading, but evidence seems to suggest you are reading wrong or reading into something else.

Group 2:
J (summarized):
There is no need to try to place the microphone at the same points used in calibration in order to confirm a room correction system's predicted results with an independent measurement system such as REW or FuzzMeasure. Instead, the proper approach is to sample random points in the same area covered by the calibration points. That way you can test both whether a RC system (a) offers a reasonable prediction of its end result and (b) results that are stable over an area.

To elaborate, here's some data showing RC predicted-vs-actual comparisons done properly:

• @chych7, post #470 (emph. added):
  • "Dirac Live was calibrated with a UMIK-1 using the tight focusing (9 point) mic pattern; the "measurement cube" length was around 2 ft. Audyssey was setup using the same measurement points with the Audyssey mic (there will be variation in exact mic positioning though) [JBH note - I don't think that's actually necessary, but can't hurt and will stave off dumb comments in forums.]...I did a 6 point measurement with REW at positions within the 9 point setup pattern
• Fig. 22 of my Denon X4100 review, "Audyssey Reference Displayed vs. Measured Inverse Curve" See text for methods.
• Fig. 23 of my Bryston SP4 review, "Dirac Live Auto-Target Fit." See text for methods.

Of note, both systems seem to be doing a reasonable job in my old setup of showing you what they do. As I wrote above, @chych7's measurements uncovered a defect in MultEQ X, so results obtained from basic Audyssey XT32 may not apply to MultEQ X. (A broken mic, if that ends up being the root cause of the defective calibration, is still an Audyssey MultEQ X defect in my view.



When you have the facts on your side, argue the facts.
When you don't have the facts on your side, argue the theory.
When you have neither the facts or theory on your side, lob condescending bullshit.

Wow, "condescending bullshit"? I have been polite and respectful, so please keep it civil if you are going to continue. I am not the OP, I simply commented that in your first couple of responses you referred to something different that the OP posted, right, wrong, or in between, that's my perception. You insisted you were talking about the same thing. Clearly there is no point keep arguing on that, so in my last post I have already said I would leave it to the OP to continue if he chooses to, that's up to him.

As I said before, part of being respectful, that you seem knowledgeable, but case in point, when I mentioned the Audyssey MRC enabled vs disabled, the way you contradicted me clearly showed you had the misconception on you part. When I provided the links for you to hear straight from Audyssey, you kept silent, but that's good for a change. Just because you posted something now that showed your knowledge in some areas, I have to spend time on gathering the "facts" and argue? What facts, and what to argue to/with, have I disagreed to any of those you posted, exactly what? Otherwise please don't impose on others to read your stuff, and argue with you, or even just to comment. I would though, if I feel there is a need and only if I want to, but right now, again, I have not disagreed with the things you posted under Group 2, yet.

Again, the only one point that I repeated more than once, is that you claimed you were talking about the same thing the OP did in your first couple responses (then you diverged into something else). You insisted you were talking about the same, and the OP did tell you it's different. Please be reminded that in my last post I have already said I would leave it, so I won't repeat that again.

 

[Datec]

Haider, you need to leave your courtroom argument tactics in the courtroom. I would suggest you go buy this software, spend the time working it doing all your own measurements and post your findings instead of telling people what to do and how to do it. That way you have 1st hand experience with it and solid "facts" to support your claims. $200 won't break the bank and you should have more then one mic to do those comparisons also given you say you have experience with all those other programs.

So far I haven't learned a dam thing from all this. That's a shame.

 

[jhaider]

I have NOT said any of those you quoted in Group 1 that you are implying, and if not please clarify.

Every quote was cited with a linkback.

I have only suggested that you were referring to something different than the OP's… I simply commented that in your first couple of responses you referred to something different that the OP posted, right, wrong, or in between, that's my perception… Again, the only one point that I repeated more than once, is that you claimed you were talking about the same thing the OP did in your first couple responses...

An awful lot of words, but still no actual answer regarding any alleged differences in scope. Perhaps because there is none.

Also, I’ve been consistent in explaining my position regarding confirmatory measurements of room correction results. Don’t try to gaslight with some alleged but unexplained “diversion.”

As I said before, part of being respectful, that you seem knowledgeable, but case in point, when I mentioned the Audyssey MRC enabled vs disabled, the way you contradicted me clearly showed you had the misconception on you part. When I provided the links for you to hear straight from Audyssey, you kept silent, but that's good for a change.

My data (which apparently you are too proud to read) showed you were wrong, so no need to belabor that. As for what Audyssey reps said or didn’t say, I don’t really care when it contradicts the data.

 

[peng]

Every quote was cited with a linkback.

Again I did not say those things you quoted in your so called Group 1, period. No idea why you would resort to creating something that appear to be quoting what I said in your so called "linkback", instead of just quoting what I said.

If you click on your "C (quoted, emph. added):", it won't show what your text below the link. But if people don't, they would think I said those things that I did not. Is that why you did it that way, I hope not as that would be totally ridiculous.. I would not even bother responding but I have to let others know you seem to be making things up, whether it is intentional or not.

An awful lot of words, but still no actual answer regarding any alleged differences in scope. Perhaps because there is none.

An awful lot of words, but still no actual answer regarding any alleged differences in scope. Perhaps because there is none.

Also, I’ve been consistent in explaining my position regarding confirmatory measurements of room correction results. Don’t try to gaslight with some alleged but unexplained “diversion.”

How many times I have to tell you I have said nothing about what you posted, explained, or whatever. I told you that you were talking about different (related, but not the same) things the OP talked about, and did take the time to explain the difference I perceived. You insisted you did not miscommunicate, that's fine, loud and clear, you should be happy now so please move on. I am not interested in playing silly games.

 

[Reverend Slim]

Now now, girls... You're both pretty.

 

[jhaider]

Again I did not say those things you quoted in your so called Group 1, period.

Agreed, in part. The quote attributed to “P” with a link back to your post is your text.

To recap, my post included the following ask: “please explain to the group the relevant and material distinctions you see between the following statements.”

Three quotes were then included, each with separate attribution and link back, to level set regarding the alleged difference in scope. I tried in good faith to keep things in context, including the “he said” in my quote from your post to make clear you were relaying your impression of someone else’s words, but the linkbacks were provided in case I fell short in that.

How many times I have to tell you I have said nothing about what you posted, explained, or whatever.

You said I was writing about something different than what C was discussing, which is certainly saying something about what I wrote! My request to you was and is to explain the distinction(s) you see, because I see none. You may have a point, but it’s hard to see through all the handwaving and gaslighting. Here’s a template for a good faith response.

“C was writing about x. You interpreted C to be writing about y. They are different, because z.”

I would suggest you go buy this software, spend the time working it doing all your own measurements and post your findings instead of telling people what to do and how to do it.

Not currently interested. Unfortunately, the feature set currently offered does not, to me, appear to add value over the $20 iOS app. See here for a deeper dive. I hope that changes as MultEQ X develops.

As for my own measurements, I trust the data posted earlier in this thread by @chych7. The methods were clearly explained, and as described they reflect best practices as I understand them. That data showed level and directional errors in MultEQ X’s target curve fit. That was surprising to me, and counter to my experience with OG MultEQ XT32. At this point we don’t know whether these errors stem from a bug or bugs in the program, or a bad mic. I hope @chych7 sends a bug report to Audyssey, so they can fix their software or send out an in-spec microphone.

 

[GalZohar]

Esteemed colleagues, I ran some tests comparing MultiEQ to Dirac and I'm thoroughly confused. I'm trying to decide between buying a full DIRAC license vs getting MultiEQ X. I did buy the ipad MultiEQ app.

I didn't have enough time today, so I didn't actually take multiple mic position measurements which I know is a problem. In any case, I thought I could learn something in the relative differences between the two if I had the mic in the same position for the tests.

For all of the measurements, I kept the mic position in the main center position I used for EQ. The room curve I am using is the Harman 6dB curve. For DIRAC setup I used a calibrated USB mic, for MultiEQ, I used the mic that came with the receiver. For the tests below, I used the same calibrated USB mic.

It's clear from below that DIRAC is producing the closest to the target room curve.
Orange - No EQ
Blue - MultiEQ reference setting
Purple - MultiEQ flat setting
Dark Red - DIRAC
View attachment 188695

I was trying to understand what the actual EQ that was being applied so I tried feeding the preamp output of my AVR into the MIC input of my sound card input and the curves below are what I measured. What the heck? MultiEQ doesn't seem to do anything below 200 Hz. And why is Flat called Flat, it's anything but Flat? It's more like kill the bass and enhance the treble, almost a reverse typical room response. Dirac is doing what it's supposed, and MultiEQ is kinda similar to Dirac above 1kHz. What could I be doing wrong in my setup? I checked my speaker config and I had them set to large with no crossover and no sub.

DIRAC - Dark Blue
MultiEQ Flat - Light Blue
MultiEQ Reference - Green
View attachment 188694

I'm going to keep testing until figure this out. I'll probably re-run the EQ setup and mark the mic locations so I can run the post EQ measurements more accurately.

This seems like XT rather than XT32? The no bass correction is a limitation of XT, which shouldn't be compared to any of the better EQ systems...


As for verification of measurement, I think if we want to test how well a system did compared to what it was supposed to do, you need to measure the exact same points. If you want to measure how well it did in general, including how well your microphone positions translate to a good spatially-averaged result, then using slightly different positions could be a good idea. The problem with different positions is that if you get bad results there, it might not be the fault of Audyssey, but rather the fault of the positions you chose to use during the calibration, so to get a fair comparison the same positions must be used. Granted as long as you use the exact same points for measurement for the 2 systems you compare, and also the exact same points of measurement to verify the results (could be different points than the first set), then that would be a fair comparison as well.

Subwoofer integration
Another question - Did anyone manage to get any improvement in crossover behavior by changing the low frequency rolloff frequency? Meaning getting any better subwoofer/speaker integration. Without messing with it, seems like default curve applies a 12db/oct rolloff which is supposedly expected to add up with the AVR's 12db/oct crossover slope, while the subwoofer gets a 24db/octave (seems like its curve has no rolloff within the corrected range, well above any practical crossover setting). However, if you increase the crossover then your slopes will not be aligned, and supposedly that could cause integration issues. I was wondering if anyone got any measurable and audible improvement through the app in that regard. Also wondering if something similar can somehow be achieved with the android app, as it doesn't seem to allow any changes to the low frequency rolloff in the curve.


Finally, isn't this thread in the wrong forum? Without google I would never have found it.

 

[Chromatischism]

Subwoofer integration
Another question - Did anyone manage to get any improvement in crossover behavior by changing the low frequency rolloff frequency? Meaning getting any better subwoofer/speaker integration. Without messing with it, seems like default curve applies a 12db/oct rolloff which is supposedly expected to add up with the AVR's 12db/oct crossover slope, while the subwoofer gets a 24db/octave (seems like its curve has no rolloff within the corrected range, well above any practical crossover setting). However, if you increase the crossover then your slopes will not be aligned, and supposedly that could cause integration issues. I was wondering if anyone got any measurable and audible improvement through the app in that regard. Also wondering if something similar can somehow be achieved with the android app, as it doesn't seem to allow any changes to the low frequency rolloff in the curve.

My understanding is that Audyssey corrects the frequency response of speakers in the absence of a crossover, so there are no problems changing crossover point outside of trying to set it lower than the detected F3 point, because there are no filters applied below it.

 

[GalZohar]

What I mean is that if it sets a 12db/oct slope based on 60Hz, but then you set the crossover to 80Hz, then you won't have a proper 24db/oct slope at 80Hz but rather just 12db/oct (with an additional 12db/oct joining in at 60Hz), while the subwoofer will get 24db/oct at 80Hz. I'm nots sure how detrimental this is to the overall result, but as far as I'm aware it's not really how it's supposed to be for optimal results. Seems like MultEQ X can change that but the app can't? I was wondering if anyone managed to use that to improve the result.

 

[Reverend Slim]

What I mean is that if it sets a 12db/oct slope based on 60Hz, but then you set the crossover to 80Hz, then you won't have a proper 24db/oct slope at 80Hz but rather just 12db/oct (with an additional 12db/oct joining in at 60Hz), while the subwoofer will get 24db/oct at 80Hz. I'm nots sure how detrimental this is to the overall result, but as far as I'm aware it's not really how it's supposed to be for optimal results. Seems like MultEQ X can change that but the app can't? I was wondering if anyone managed to use that to improve the result.

I've been fiddling with this, but I'm a little confused on how the cutoff override interacts with the low frequency EQ limit in the next step. For instance, my heights (SVS Prime Elevations) and surrounds (SVS Prime Satellites) exhibit a rolloff that pretty closely matches 4th order... but they're being detected as 2nd. So I've been overriding them on Design Target Curve to make them 4th, comparing to how they're actually rolling off on the measurement screen. Then on the Filter Settings, you can see that 4th order rolloff being applied using the filters, but if you change to manual and put any other frequency to stop filtering, your rolloff goes away. Even if you use the same detected number (i.e. Auto 80 changed to Manual 80), the rolloff disappears from the chart. So I'm assuming leaving them on Auto is the only way to keep the cutoffs you assigned under Design Target Curve. I've had pretty good results from changing my heights and surrounds to 4th order though.

Contrast that with my mains and center (Prime Towers and Center), which get detected as 2nd order at 30Hz (Large) and 40Hz. I run these with 40 and 60Hz crossovers, which gave me the best integration. Then I tried changing them to 4th order as well and it seemed to make bass in those channels sound more punchy. I'm not really sure why that would be, since those rolloffs would be below the crossover points I'm assigning. But that is what has given me the best sound.

So I'm not really sure how the cutoffs integrate with the AVR's crossover. Of course, I'm sure other here know more about this than me, so if anyone has any suggestions based on everything said here, feel free to toss 'em at me.

 

[chych7]

Alright folks, I think I've found evidence for one area where MultEQ-X is better than the App - time alignment. MultEQ-X computes different distances than the App, as many of us have noticed, so that suggested to me a possible difference in time alignment.

It took me a bit to get multichannel measurements working in REW (the current latest version does not work with ASIO4All drivers, had to use v5.17). Only the 7 floor channels can be accessed this way, so Atmos channels could not be measured. Here I set up the calibration with just 7 speakers, no sub or Atmos ceiling (I nominally have a 7.2.4 system), using the default Audyssey measurement positions, default target curve, and MRC off. I measured 500 Hz - 20 kHz (I wanted to avoid room modes in this analysis) with REW with acoustic timing reference (right channel), and report below the delay values computed. Ideally they should all be identical.

	App $20	MultEQ-X $200
Left	0.40	0.40
Right	0.31	0.31
Center	0.50	0.50
Back Left	0.81	0.60
Back Right	0.71	0.46
Surround Left	1.1	0.60
Surround Right	0.85	0.35

Note: units of measurement are milliseconds

LRC are matched between the two calibrations, however the surround speakers are much better aligned with MultEQ-X. I would assume that the Atmos speakers are also better time aligned, but I don't think it's possible to access those channels in REW to confirm it. Better time alignment should present itself as improved imaging/spatial positioning, which I think agrees with what some people have subjectively noted. I'm not sure how audible a ~0.5 ms shift in delay is, but it would be a significant phase shift for mid-high frequencies, and we can hear phase mismatches.

Now, do you need the $200 app for better time alignment? You could do it with REW and a lot of painful calibration for the floor speakers, but as far as I know, you can't with the Atmos speakers.

SPL vs. frequency plots are below, they look very similar across the two calibrations.

App

MultEQ-X

If anyone else can repeat this analysis, it would help build up some statistical confidence.

 

[GalZohar]

It is strange if you got better time alignment due to multeq x rather than just luck or better mic placement, isn't it? Seems like the measurements are still 100% in the AVR, which should be the same.

I am wondering about the rolloffs and crossovers, because that is the only significant difference I found compared to ARC or Dirac (as far as I understand). The overall EQ per channel (crossovers aside) seems similar based on what others on the forums are getting. But I want to know if I am missing something with subwoofer integration with multeq and if there is a good way to fix it with X or not.

 

[Reverend Slim]

It is strange if you got better time alignment due to multeq x rather than just luck or better mic placement, isn't it? Seems like the measurements are still 100% in the AVR, which should be the same.

While the measurements are still in the AVR, I think it's in the assignment it makes to the AVR that is the difference, which is why distances on the AVR no longer match the detected distances on the app (even when, in my case, the distances actually detected between the two are pretty close). My assumption was that they're using a lookup table to see what distances in the AVR get the differential delays closer. In other words, the app doesn't care what the detected distance is so much as the difference in delay between each channel, and then picks distances in the AVR that get the cross-channel delays closer to ideal. That's certainly something they could do in the phone app as well, but as they said in the video, it was something they only discovered they could do when making the PC app. Maybe it will work its way downward into the phone app... or maybe they'll leave it as a "feature" just for MultEQ X.

 

[Fidji]

	App $20	MultEQ-X $200
Left	0.40	0.40
Right	0.31	0.31
Center	0.50	0.50
Back Left	0.81	0.60
Back Right	0.71	0.46
Surround Left	1.1	0.60
Surround Right	0.85	0.35

Note: units of measurement are milliseconds

Both App and Multeq X are doing the same mistake in my set-up [I have stationary microphone stand, so I am able to repeat measurements]. This is why I always adjsut L-C-R Manually [3cm L, approx 7 cm C] until I get proper REW impulse response in MLP. I do get pretty consistent delays between App and MQX

If you want to measure Atmos, you just need to re-assign amps [no idea whether your AVR allows it, it is possible for 8500] and then measure.

I would not be hang-up on SR/SRB impulses - they will be completely different for each of the listening positions.

No offence intended - your Center FR really looks ugly. Is this after MQX has done its miracles?

 

[chych7]

Both App and Multeq X are doing the same mistake in my set-up [I have stationary microphone stand, so I am able to repeat measurements]. This is why I always adjsut L-C-R Manually [3cm L, approx 7 cm C] until I get proper REW impulse response in MLP. I do get pretty consistent delays between App and MQX

Can you elaborate on your method of adjusting distances/reference a guide? I haven't looked into doing that for my fronts.

If you want to measure Atmos, you just need to re-assign amps [no idea whether your AVR allows it, it is possible for 8500] and then measure.

No I can't do this with my receiver.

I would not be hang-up on SR/SRB impulses - they will be completely different for each of the listening positions.

Yeah now that I think about it, 1 ft change is a 0.9 ms shift, so such measurements would be very sensitive to position. I'll have to redo this and see what the results look like at different positions, to see if that changes the conclusion.

No offence intended - your Center FR really looks ugly. Is this after MQX has done its miracles?

Yeah my center looks wonky, I think it's an interaction with my seating, and the speaker being positioned below the screen. I really should be averaging the measurement across multiple points. It measures fairly flat near-field.

 

[Fidji]

Can you elaborate on your method of adjusting distances/reference a guide? I haven't looked into doing that for my fronts.


Yeah my center looks wonky, I think it's an interaction with my seating, and the speaker being positioned below the screen. I really should be averaging the measurement across multiple points. It measures fairly flat near-field

I do what you do - measure the impulse response (using center for timing reference) and then manually adjust distances in MQX. Afterwards I re-check in MLP for L-C-R.

Actually I do focus much more on getting the gain right - using 1/1 or 1/2 octave smoothing in REW to get overall picture.

 

[anotherhobby]

This weekend I bought MultEQ-X on a whim, set it up on my Mac, and sat down with it for a day to see how it well it worked for my setup. I'll share some thoughts and measurements below later in a different post, but for now I just wanted to mention this tip for Mac and Linux users: You can run MultEQ-X without the added expense of virtualization software or having to pay for Windows (but you still need to deal with running Windows in virtualization software).

Just download VirtualBox (which is free) and the Windows10 disk image directly from Microsoft (which is free). Install Virtual Box and then install Windows10 on top. You can legally use Windows10 for free if you don't care about customizing/personalizing it at all. That'd suck if you wanted to use it as your daily OS, but for this case you don't. It is also slightly annoying that you'll need to create a Microsoft account to buy MultEQ-X thru their store, but it all works just fine. I had zero problems or hiccups at all. This also works great for running Multi-Sub Optimizer, which is also Windows only.

There are plenty of guides around for how to setup VirtualBox and Windows. Just make sure you give VirtualBox plenty of memory (I recommend 8 GB) and CPU (I gave it 4 cores on my 2018 MacBook Pro). If you starve Windows of memory or CPU the software will run very slow.