• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Audyssey Room EQ Review

Reverend Slim

Member
Joined
Feb 18, 2022
Messages
76
Likes
58
Location
Mobile, AL
According to Audyssey/D+M (in the recent YT video), the dip shown really isn't the target as such (strictly speaking, as it actually depends..). Whether you will end up with that dip or not will depend on your speakers. It the speakers are not designed to have that dip then you will not have it even if you leave it enabled. If the speaker is designed with the dip then it will have the dip unless you disable it, in that case Audyssey would try to flatten the dip. In the video, the Denon rep suggested to leave it enabled so as not to change the speaker's design, and disabled only if you tried, and don't prefer the sound with the dip.
I don't think you're quite right here. If you leave MRC enabled, irrespective of whether your speaker exhibits that dip or not, Audyssey uses its filters to craft its MRC dip centered around 2kHz in the target curve. It doesn't do any detection or analysis of whether your speaker already has that dip, so it couldn't possibly ignore it if it does. But if the measured response has that dip already at that particular frequency, naturally it wouldn't need to use any filters to achieve it (or at least minimal filtering in that region to match the target). My ear-level speakers don't exhibit that dip in my room (though their crossover is at 2.5kHz, so any dip they had wouldn't match Audyssey's MRC anyway), but if I load two profiles to my AVR with MultEQ X - one with MRC and one without - there's an obvious difference. If what you're saying is true, there wouldn't be.

My height channels, however, do exhibit a dip at 2.5kHz due to them being more off-axis than I'd like. If I enable MRC for those channels, Audyssey still uses its filters to craft a dip at 2kHz. The existing measured dip that's at 2.5kHz gets flattened out (or at least attempts to). Audyssey has said that you can disable MRC and manually create your own version of the filter at the crossover region of your particular speaker... but there's no fast way to do it. They have suggested that a future update for MultEQ X may allow you to enter in your own center frequency for their MRC, which would let you better tailor it to your individual speakers. Of course, that is with MultEQ X. With the app, you can't create your own matching filter - just disable/enable MRC in the target curve.

Short version: If you see a dip in measured response around 2kHz, leave MRC on for that speaker so it doesn't try to fight it with filters. If you don't, try MRC on and off to see which you prefer.
 

krabapple

Major Contributor
Forum Donor
Joined
Apr 15, 2016
Messages
3,169
Likes
3,717
I don't think you're quite right here. If you leave MRC enabled, irrespective of whether your speaker exhibits that dip or not, Audyssey uses its filters to craft its MRC dip centered around 2kHz in the target curve. It doesn't do any detection or analysis of whether your speaker already has that dip, so it couldn't possibly ignore it if it does. But if the measured response has that dip already at that particular frequency, naturally it wouldn't need to use any filters to achieve it (or at least minimal filtering in that region to match the target). My ear-level speakers don't exhibit that dip in my room (though their crossover is at 2.5kHz, so any dip they had wouldn't match Audyssey's MRC anyway), but if I load two profiles to my AVR with MultEQ X - one with MRC and one without - there's an obvious difference. If what you're saying is true, there wouldn't be.

Does you system allow instant switching between profiles? Mine doesn't, and the time lag to load them would preclude confidence in sighted A/B comparisons, even if the result seems 'obvious'.
 

peng

Master Contributor
Forum Donor
Joined
May 12, 2019
Messages
5,615
Likes
5,168
I don't think you're quite right here. If you leave MRC enabled, irrespective of whether your speaker exhibits that dip or not, Audyssey uses its filters to craft its MRC dip centered around 2kHz in the target curve. It doesn't do any detection or analysis of whether your speaker already has that dip, so it couldn't possibly ignore it if it does. But if the measured response has that dip already at that particular frequency, naturally it wouldn't need to use any filters to achieve it (or at least minimal filtering in that region to match the target).

Of course, I may be wrong, or at least as you said, "not quite right"... The first time I mentioned this MRC thing, iirc., I included the link to that Audyssey/D+M video. In fact, in my previous posts, I said that people, including me, misunderstood the MRC disabled vs enabled thing but that's again I must emphasized what I cited was based on what Audyssey/D+M said in the linked video. To void being taken as spreading misinformation, I referenced the video, posted the link and, even took to time find and referenced the time mark so there would be no need to watch the whole video if they didn't want to. I rarely get annoyed by forum posts, but in this case someone accused me of repeating misinformation when I actually tried to highlight the fact that the claims of some people about Audyssey "creates" the MRC dip around 2 kHz regardless of the speaker's own FR at around 2 kHz may actually be based on incorrect interpretation of the whole MRC enabled/disabled thing. If anyone want to dispute what Audyssey/D+M said in the reference video, they should contact Audyssey or Phil Jones of Denon. I would not dispute it myself because my own findings seem to confirm what they said. Also, as I also quoted Audyssey, that people should look at their speaker's measured response to see if they have such midrange dips, and go from there. So one does not need to blindly choose whether to leave MRC enabled of disabled. Or if one want it disabled regardless, that's their decision, but just realize that according to that video, MRC enabled means if the speaker has a dip around 2 kHz, Audyssey would leave it there without trying to EQ it out, but Audyssey would not create such a dip from nowhere. Again, it's not what I said, its what I heard/understood in that video.

My ear-level speakers don't exhibit that dip in my room (though their crossover is at 2.5kHz, so any dip they had wouldn't match Audyssey's MRC anyway), but if I load two profiles to my AVR with MultEQ X - one with MRC and one without - there's an obvious difference. If what you're saying is true, there wouldn't be.

Yes, I noticed the same in some of my graphs too, but it does not mean what Audyssey said (again, please realize what I said was based on what Jeff and Phil said in the video, or at least my interpretation..) was not "true". Even if your curve shows a dip at around 2,000 Hz, it does not mean Audyssey created it, it could still be how your speaker's response is in your room. My Philharmonic BMR shows a dip around there too, without any help from Audyssey. As we know, placement, room do influence the midrange response too. So it could still be normal to have the two curves to look different whether your speakers have a dip in centered around the XO point at 2500 Hz or not.

My height channels, however, do exhibit a dip at 2.5kHz due to them being more off-axis than I'd like. If I enable MRC for those channels, Audyssey still uses its filters to craft a dip at 2kHz.

As I tried to explain above, that does not mean Audyssey still would "uses its filters to craft a dip at 2 kHz..." I have used Audyssey with several of my speakers and I have seen no proof.. (Edit: It could be for other reasons that I didn't see any obvious Audyssey created dips, but I believe you are right that Audyssey would do it as shown in their target curve.) If, in addition the the 2.5 kHz dip, you also have one at around 2.0 kHz, it could just be the way your speaker responded in the room and in the way it is placed.

Short version: If you see a dip in measured response around 2kHz, leave MRC on for that speaker so it doesn't try to fight it with filters. If you don't, try MRC on and off to see which you prefer.

Here you seem to be saying what I am saying, so obviously I agree...:)

I'll link the video again here so you can listen to that part yourself if you are interested. As I said, I did look at some of my REW graphs, and to me they do seem to indicate what Jeff and Phil (the Audyssey and D+M rep) said seem to explain what I saw in my own graphs.


From time: 54:30 to about 1:00, the important part of the conversation starts at about 56:00. They included two speaker's curves, one of them shows the speaker has the dip at about 3,500 Hz. So in that case, I suppose leaving MRC enabled would not make a noticeable difference.
 
Last edited:

Reverend Slim

Member
Joined
Feb 18, 2022
Messages
76
Likes
58
Location
Mobile, AL
Does you system allow instant switching between profiles? Mine doesn't, and the time lag to load them would preclude confidence in sighted A/B comparisons, even if the result seems 'obvious'.
As fast as changing between reference and flat, which is where MultEQ X loads them. With MultEQ X, I can save any combo of targets to those locations. I've used this feature to quickly A/B different tweaks as I make them.
 

peng

Master Contributor
Forum Donor
Joined
May 12, 2019
Messages
5,615
Likes
5,168
My ear-level speakers don't exhibit that dip in my room (though their crossover is at 2.5kHz, so any dip they had wouldn't match Audyssey's MRC anyway), but if I load two profiles to my AVR with MultEQ X - one with MRC and one without - there's an obvious difference. If what you're saying is true, there wouldn't be.

I forget to mention that instead of comparing the two profiles, one with MRC enabled and the other disabled, you could compare the one with MRC enabled and Audyssey Off. That, in my opinion, would be a better comparison, to see if Audyssey would really "craft" a dip at 2 kHz regardless. Still it is not going to be 100%, but may be 90% better and easier to see the effects.

Here are two I could find quickly:
It was from using my AV8801 demoted to my secondary two channel system, so no App, that is MRC "on", or "enabled" naturally..

1651088940179.jpeg


1651088304872.jpeg
 

Reverend Slim

Member
Joined
Feb 18, 2022
Messages
76
Likes
58
Location
Mobile, AL
I rarely get annoyed by forum posts, but in this case someone accused me of repeating misinformation when I actually tried to highlight the fact that the claims of some people about Audyssey "creates" the MRC dip around 2 kHz regardless of the speaker's own FR at around 2 kHz may actually be based on incorrect interpretation of the whole MRC enabled/disabled thing.
I don't think you're "repeating misinformation" so much as slightly misinterpreting what is said in that video. Audyssey technically does "create" that MRC dip regardless of the speaker's FR around 2kHz if MRC is enabled. I say that because its filtering is done to Audyssey's pre-defined MRC dip, which may well differ in level, frequency and Q from what an individual speaker has implemented in its crossover from mid to tweeter. So technically, if your speaker doesn't have that dip and MRC is engaged, Audyssey creates it using filters such that the resulting target is flat except for that dip (and any defined high end rolloff, naturally). If your speaker DOES have that dip at the exact same frequency, level and Q, then Audyssey just doesn't use any filters in that frequency range... so the result is that the dip still exists post-EQ. If your speaker has a dip at a different frequency, Audyssey attempts to flatten that dip using filters and then create its own pre-defined dip centered at 2kHz using filters.
Also, as I also quoted Audyssey, that people should look at their speaker's measured response to see if they have such midrange dips, and go from there. So one does not need to blindly choose whether to leave MRC enabled of disabled. Or if one want it disabled regardless, that's their decision, but just realize that according to that video, MRC enabled means if the speaker has a dip around 2 kHz, Audyssey would leave it there without trying to EQ it out, but Audyssey would not create such a dip from nowhere. Again, it's not what I said, its what I heard/understood in that video.
That's the part you're misunderstanding. Yes, if the speaker has a dip around that region, Audyssey doesn't have to do any filtering to try to match the target curve. But that doesn't mean that if a speaker has flat response across that region instead of a dip, Audyssey just ignores MRC. It is still going to try to apply filtering to make the response conform to the target curve, which if you have MRC engaged, includes that dip centered at 2kHz. If you go to about 1:04:22 in that video, you can see the Filter Settings screen. What you see there is the inverse filter that Audyssey is applying to counter the measured response such that it reaches the intended target curve. So if you look at the LEFT speaker at that time stamp, you can see the BEFORE line in green that shows a broad dip in response around 3.5kHz. You can also see the inverse filter applied (in purple) to counter that. Now look at the YELLOW line, which is the theoretical post-EQ response based on the filtering. If you look at 2kHz, you'll see the MRC dip. It isn't there on the BEFORE measurement in any way. Audyssey created it with filters to match the intended target response (which you can see includes the MRC dip if you scroll back to 1:01:22).
From time: 54:30 to about 1:00, the important part of the conversation starts at about 56:00. They included two speaker's curves, one of them shows the speaker has the dip at about 3,500 Hz. So in that case, I suppose leaving MRC enabled would not make a noticeable difference.
But again... scroll forward to where they show the Filter Settings screen and you'll see that the wide-Q dip at 3.5kHz has been EQ'd flat, with the only expected dip remaining post-EQ being the narrower MRC at 2kHz. It is essentially removing the speaker's natural dip across the crossover range and applying its own dip at a different frequency.

Two things worth mentioning having said all that:
1. Some of this is why people, whether on the app or MultEQ X, apply a curtain so that Audyssey applies no filtering whatsoever above the Schroeder frequency. Doing so would inherently ignore whether MRC is on or off, because you're not applying ANY filters in the 2kHz region at that point.
2. The big caveat: Just because Audyssey believes that yellow line should be the post-EQ response, that doesn't necessarily mean that's what you'll measure in the room. It will certainly be closer to that theoretical end result than it is to the original measured response... but even if Audyssey's filters try to create that MRC dip, you may still not measure that exact post-EQ dip after all is said and done.
 

Reverend Slim

Member
Joined
Feb 18, 2022
Messages
76
Likes
58
Location
Mobile, AL
I forget to mention that instead of comparing the two profiles, one with MRC enabled and the other disabled, you could compare the one with MRC enabled and Audyssey Off. That, in my opinion, would be a better comparison, to see if Audyssey would really "craft" a dip at 2 kHz regardless. Still it is not going to be 100%, but may be 90% better and easier to see the effects.
If you want to really get confused... If application of the MRC dip in Audyssey's filters results in the average overall level of a particular speaker no longer matching the rest of the speakers, it will actually shift the entire inverse filter such that the average level stays consistent. You won't typically see it on full-range speakers, but if you see the filters on speakers that tend to roll off around 80-100Hz (like my Prime Elevations), the entire target line shifts so that the overall level is raised without the coarse level trim having to change between the two states (i.e. if you have one profile with MRC on and one with MRC off).

If I were at home right now, I could show you what I'm talking about, but I hope that made sense.
 

bjmsam

Member
Joined
Feb 1, 2022
Messages
41
Likes
21
Location
Mount Airy, MD
I forget to mention that instead of comparing the two profiles, one with MRC enabled and the other disabled, you could compare the one with MRC enabled and Audyssey Off. That, in my opinion, would be a better comparison, to see if Audyssey would really "craft" a dip at 2 kHz regardless. Still it is not going to be 100%, but may be 90% better and easier to see the effects.
That is what the graph I posted depicts. Audyssey ON (with MRC) pulls it down and Audyssey FLAT (without MRC) pushes it up.
 

peng

Master Contributor
Forum Donor
Joined
May 12, 2019
Messages
5,615
Likes
5,168
That is what the graph I posted depicts. Audyssey ON (with MRC) pulls it down and Audyssey FLAT (without MRC) pushes it up.

I looked at the ones you posted and noticed that in relative term I couldn't identify the kind of dip at 2 kHz that resemble the Audyssey target curve so I really don't know what to say...
 
Last edited:

Reverend Slim

Member
Joined
Feb 18, 2022
Messages
76
Likes
58
Location
Mobile, AL
Peng, I'm stealing one of your graphs from another thread to illustrate what I'm talking about.
1640993412227.jpeg

Look at the response at 2kHz. With no equalization (red), you see that the speaker has a bit of a dip, though it actually seems to be centered more around 2.3kHz and it's hard to say if that's an actual dip since response seems to generally tilt downward. With Audyssey Reference (black, which uses MRC), that dip is maintained but made more narrow and you can see that everything above about 2.5kHz has been eq'd up. With Audyssey Flat (blue, which does not have MRC), you can see that it has EQ'd the dip out (and see above 7kHz that it isn't rolling off the highs).

We're not talking about a huge difference here.
 

Dennis_FL

Addicted to Fun and Learning
Forum Donor
Joined
Feb 21, 2020
Messages
525
Likes
413

Chromatischism

Major Contributor
Forum Donor
Joined
Jun 5, 2020
Messages
4,767
Likes
3,706
Not sure if it’s been mentioned, but in the latest version of the app (downloaded last week as I now have a Denon avr) you have an option for midrange compensation which turns off the bbc dip in the target curve
It's not a BBC dip. It's a dip in the sound power at the crossover. All non-coaxial speakers have them. If your speakers are coaxial or they engineered around it, turn it off. If your speakers have such a dip but it occurs at a different frequency than stock, turn it off and draw your own on the curve editor.
 

peng

Master Contributor
Forum Donor
Joined
May 12, 2019
Messages
5,615
Likes
5,168
I don't think you're "repeating misinformation" so much as slightly misinterpreting what is said in that video. Audyssey technically does "create" that MRC dip regardless of the speaker's FR around 2kHz if MRC is enabled.

In that case, I would say one or both of us may be misinterpreting something they said. Edit: Now I believe you are right, at least technically.

I say that because its filtering is done to Audyssey's pre-defined MRC dip, which may well differ in level, frequency and Q from what an individual speaker has implemented in its crossover from mid to tweeter. So technically, if your speaker doesn't have that dip and MRC is engaged, Audyssey creates it using filters such that the resulting target is flat except for that dip (and any defined high end rolloff, naturally). If your speaker DOES have that dip at the exact same frequency, level and Q, then Audyssey just doesn't use any filters in that frequency range... so the result is that the dip still exists post-EQ. If your speaker has a dip at a different frequency, Audyssey attempts to flatten that dip using filters and then create its own pre-defined dip centered at 2kHz using filters.

I fully understanding what you are saying, but I don't think that's what those two gentlemen are saying. Yes they show that dip in the target curve and that's why I said they were the one who created the misunderstanding. You are still assuming that because they show it in the target curve, then Audyssey will create the dip accordingly unless the measured response has the exact level, frequency and Q and that's absolutely normal to assume that because of the way Audyssey presented it in their target curves. As you know, my interpretation is not just slightly different than yours, but almost directly opposite. My understanding is that if you leave MRC enabled then Audyssey will simply not EQ (to flat) any dip found where the target curve shows, and if you disable it, then it will try to EQ it (dip or not..) to flat. Thanks to you, I am not longer clear whether you, or I , or both are misinterpreting.., just not sure any morew. I may email Audyssey about this and if I do, I will certainly let you know the outcome.

That's the part you're misunderstanding. Yes, if the speaker has a dip around that region, Audyssey doesn't have to do any filtering to try to match the target curve. But that doesn't mean that if a speaker has flat response across that region instead of a dip, Audyssey just ignores MRC. It is still going to try to apply filtering to make the response conform to the target curve, which if you have MRC engaged, includes that dip centered at 2kHz.

Same deal, not sure who's misunderstanding though lol.. You are assuming Audyssey is still going to try to apply filtering to enforce the dip per target curve even if the measured response is flat, or even a hump? That's your understanding, but as you know, not mine so please don't assume mine is the "misunderstanding". I would agree to, we are not sure, or at least I am no longer so sure and I thank you for making me think..

If you go to about 1:04:22 in that video, you can see the Filter Settings screen. What you see there is the inverse filter that Audyssey is applying to counter the measured response such that it reaches the intended target curve. So if you look at the LEFT speaker at that time stamp, you can see the BEFORE line in green that shows a broad dip in response around 3.5kHz. You can also see the inverse filter applied (in purple) to counter that. Now look at the YELLOW line, which is the theoretical post-EQ response based on the filtering. If you look at 2kHz, you'll see the MRC dip. It isn't there on the BEFORE measurement in any way. Audyssey created it with filters to match the intended target response (which you can see includes the MRC dip if you scroll back to 1:01:22).

I did look there and I could see the before (green) curve did have a dip, so it should expected that the after (yellow) curve would have a dip and I could see that the filter line did try to push things up from around there. That is not a good example or proof, to show Audyssey created the MRC dip. May be I missed something, or looked at the wrong part? I just didn't see what you saw.


But again... scroll forward to where they show the Filter Settings screen and you'll see that the wide-Q dip at 3.5kHz has been EQ'd flat, with the only expected dip remaining post-EQ being the narrower MRC at 2kHz. It is essentially removing the speaker's natural dip across the crossover range and applying its own dip at a different frequency.

Removing the speaker's natural dip across the XO range yes, applying its own dip at a different frequency I don't think so, again may be I missed something?
Two things worth mentioning having said all that:
1. Some of this is why people, whether on the app or MultEQ X, apply a curtain so that Audyssey applies no filtering whatsoever above the Schroeder frequency. Doing so would inherently ignore whether MRC is on or off, because you're not applying ANY filters in the 2kHz region at that point.
2. The big caveat: Just because Audyssey believes that yellow line should be the post-EQ response, that doesn't necessarily mean that's what you'll measure in the room. It will certainly be closer to that theoretical end result than it is to the original measured response... but even if Audyssey's filters try to create that MRC dip, you may still not measure that exact post-EQ dip after all is said and done.

I can agree to all of the above, and could have said almost the same you did.
 
Last edited:

peng

Master Contributor
Forum Donor
Joined
May 12, 2019
Messages
5,615
Likes
5,168
Peng, I'm stealing one of your graphs from another thread to illustrate what I'm talking about.
View attachment 202922
Look at the response at 2kHz. With no equalization (red), you see that the speaker has a bit of a dip, though it actually seems to be centered more around 2.3kHz and it's hard to say if that's an actual dip since response seems to generally tilt downward. With Audyssey Reference (black, which uses MRC), that dip is maintained but made more narrow and you can see that everything above about 2.5kHz has been eq'd up. With Audyssey Flat (blue, which does not have MRC), you can see that it has EQ'd the dip out (and see above 7kHz that it isn't rolling off the highs).

We're not talking about a huge difference here.

As you said, the difference there were not huge. Also, I would not have picked that graph because it might have been done after I started using the App, so all bets are off. That's why I picked the ones posted in post#705, because then I know for sure the reference and flat curves would have MRC enabled (no option). You don't see any significant difference between the ref vs off there at around 2K Hz.
 

GalZohar

Senior Member
Joined
Aug 8, 2021
Messages
442
Likes
216
As far as I understand the midrange compensation dip is just another modification to the target curve, and Audyssey will try to EQ the response to match that curve according to its algorithms, and there supposedly shouldn't be any difference between how it EQs things to that dip compared to how it EQs things to match any other modifications made to the target curve. Basically, seems like if you want to keep some non-flat features of your measured frequency response then you should modify your target curve to represent that, and otherwise keep it flat (or tilted). Or just apply a curtain and not worry about EQing high frequencies at all - In the end you should probably always compare a frequency-limited correction to your best-sounding full-range curve, as no matter how much you tweaked that curve to sound optimal, it still might sound worse than no EQ, and there are many posts on these forums explaining why.
 

Reverend Slim

Member
Joined
Feb 18, 2022
Messages
76
Likes
58
Location
Mobile, AL
You are still assuming that because they show it in the target curve, then Audyssey will create the dip accordingly unless the measured response has the exact level, frequency and Q and that's absolutely normal to assume that because of the way Audyssey presented it in their target curves. As you know, my interpretation is not just slightly different than yours, but almost directly opposite. My understanding is that if you leave MRC enabled then Audyssey will simply not EQ (to flat) any dip found where the target curve shows, and if you disable it, then it will try to EQ it (dip or not..) to flat.
I'm not assuming anything. That's quite literally how Audyssey works. It measures the response of your speaker, then uses an inverse filter to make it match its target curve. If you have MRC on, that target curve then includes that dip at 2kHz. The system is completely unaware of whether your speaker is supposed to have a dip or not and no analysis of that occurs (because how would it know?). It merely applies filters to try to correct the measured response so that it matches the intended target curve. So naturally, if your speaker has a dip that lines up with the dip MRC puts in the target curve, it is inherently having to apply less equalization in that region to make the end response match the intended target. This isn't really as complicated as you seem to think it is. But it doesn't NOT EQ in that region as if you had applied a filter curtain to that region. In fact, if your speaker has say a -4dB dip in that region but the MRC dip is only about -2dB, it will still try to EQ that region up 2dB by way of changing the inverse filter applied.

Again, look at the video. You can see the inverse filter on the screen. If the inverse filter exactly matches the measured response, you get flat response. But in the MRC region, you can clearly see that the inverse filter is different to account for that region on the intended target curve. If that video is not clear enough, I'll be glad to screengrab MRC on and off versions of the same project so you can see how the inverse filter changes to apply that dip.
Same deal, not sure who's misunderstanding though lol.. You are assuming Audyssey is still going to try to apply filtering to enforce the dip per target curve even if the measured response is flat, or even a hump? That's your understanding, but as you know, not mine so please don't assume mine is the "misunderstanding". I would agree to, we are not sure, or at least I am no longer so sure and I thank you for making me think..
For my part, I wrote a fair chunk of the Audyssey guide over at AVS back in the day and exchanged many e-mails with Chris Kyriakakis about it. I'm pretty familiar with it. The inverse filter created is to match the target curve. That is true whether you engage MRC or make your own manual changes to the target curve, and the system is completely agnostic to what the original measured response is. The only part of the workflow wherein the original response is considered as far as where not to filter is in the aggregate response post-measurement, to determine each speaker's rolloff and the order at which it should occur (which you can now override with MultEQ X if you choose, though I think you could with Pro before this). Literally all MRC is (and I believe Jeff confirmed this in another video) is their generalized filter that you can recreate yourself at any other frequency if you choose. That's why they have talked about updating the MEQX software to allow adjustment to the center frequency of that dip, so you can tailor it to where that dip naturally occurs with your particular speaker if you want.
I did look there and I could see the before (green) curve did have a dip, so it should expected that the after (yellow) curve would have a dip and I could see that the filter line did try to push things up from around there. That is not a good example or proof, to show Audyssey created the MRC dip. May be I missed something, or looked at the wrong part? I just didn't see what you saw.
Look again at the filter line (purple). Again, to EQ flat, the before line and filter line would be exactly inverted. Is the purple line the exact inverse of the before line in the MRC region (2kHz)? NOPE. Because the filtering in that region has been modified to implement the MRC dip in the target curve. If MRC was off, the two lines would be exact inverse in that region.
 

peng

Master Contributor
Forum Donor
Joined
May 12, 2019
Messages
5,615
Likes
5,168
I'm not assuming anything. That's quite literally how Audyssey works. It measures the response of your speaker, then uses an inverse filter to make it match its target curve. If you have MRC on, that target curve then includes that dip at 2kHz. The system is completely unaware of whether your speaker is supposed to have a dip or not and no analysis of that occurs (because how would it know?). It merely applies filters to try to correct the measured response so that it matches the intended target curve. So naturally, if your speaker has a dip that lines up with the dip MRC puts in the target curve, it is inherently having to apply less equalization in that region to make the end response match the intended target. This isn't really as complicated as you seem to think it is. But it doesn't NOT EQ in that region as if you had applied a filter curtain to that region. In fact, if your speaker has say a -4dB dip in that region but the MRC dip is only about -2dB, it will still try to EQ that region up 2dB by way of changing the inverse filter applied.

Again, look at the video. You can see the inverse filter on the screen. If the inverse filter exactly matches the measured response, you get flat response. But in the MRC region, you can clearly see that the inverse filter is different to account for that region on the intended target curve. If that video is not clear enough, I'll be glad to screengrab MRC on and off versions of the same project so you can see how the inverse filter changes to apply that dip.

For my part, I wrote a fair chunk of the Audyssey guide over at AVS back in the day and exchanged many e-mails with Chris Kyriakakis about it. I'm pretty familiar with it. The inverse filter created is to match the target curve. That is true whether you engage MRC or make your own manual changes to the target curve, and the system is completely agnostic to what the original measured response is. The only part of the workflow wherein the original response is considered as far as where not to filter is in the aggregate response post-measurement, to determine each speaker's rolloff and the order at which it should occur (which you can now override with MultEQ X if you choose, though I think you could with Pro before this). Literally all MRC is (and I believe Jeff confirmed this in another video) is their generalized filter that you can recreate yourself at any other frequency if you choose. That's why they have talked about updating the MEQX software to allow adjustment to the center frequency of that dip, so you can tailor it to where that dip naturally occurs with your particular speaker if you want.

Look again at the filter line (purple). Again, to EQ flat, the before line and filter line would be exactly inverted. Is the purple line the exact inverse of the before line in the MRC region (2kHz)? NOPE. Because the filtering in that region has been modified to implement the MRC dip in the target curve. If MRC was off, the two lines would be exact inverse in that region.

I guess you are right about that's how Audyssey works, except the MRC enabled/disabled is the only grey area that I see where it may not just be about applying filters to match target in a straight forward way, because the MRC enable/disable could have been used as a "condition", that may affect how things are done in that area, versus the normal way done outside of that area. That's the only part I have in mind when suggesting you are assuming it one way, that way you interpret it and that the way I interpreted it is wrong. I would concede that I now think you are more likely correct and I am more likely to have interpreted things incorrectly. I will still try to get an answer from Audyssey. Personally, I think it is more logical to use the enable/disable switch in the way I thought it would be as it doesn't make sense to enforce such a dip if even the speaker's natural response does not have such a dip. If it has, then okay, let the user decide to preserve it or not and in that case, also let the user choose the frequency.
 

Reverend Slim

Member
Joined
Feb 18, 2022
Messages
76
Likes
58
Location
Mobile, AL
I guess you are right about that's how Audyssey works, except the MRC enabled/disabled is the only grey area that I see where it may not just be about applying filters to match target in a straight forward way, because the MRC enable/disable could have been used as a "condition", that may affect how things are done in that area, versus the normal way done outside of that area. That's the only part I have in mind when suggesting you are assuming it one way, that way you interpret it and that the way I interpreted it is wrong. I would concede that I now think you are more likely correct and I am more likely to have interpreted things incorrectly. I will still try to get an answer from Audyssey. Personally, I think it is more logical to use the enable/disable switch in the way I thought it would be as it doesn't make sense to enforce such a dip if even the speaker's natural response does not have such a dip. If it has, then okay, let the user decide to preserve it or not and in that case, also let the user choose the frequency.
Letting the end user choose the frequency is exactly what they've talked about adding (though I can't recall where I saw it). For now, you can handle it yourself by crafting your own filter modification to match the speaker's natural response in that region if you want to keep just that part of the response mostly unaltered (and this would essentially normalize differences in that dip between speaker locations due to in-room response). And it isn't "enforcing a dip" necessarily. Your speaker might actually have directionality issues in that range that aren't visible in the room due to in-room response, and putting that dip back in might reduce harshness in the vocal range in that case. All of this just gives you the flexibility to tweak things as needed.

If you want a bit more info on MRC and other filter modifications being applied, here's Jeff talking more about it around the 45 minute mark:
 

peng

Master Contributor
Forum Donor
Joined
May 12, 2019
Messages
5,615
Likes
5,168
Letting the end user choose the frequency is exactly what they've talked about adding (though I can't recall where I saw it). For now, you can handle it yourself by crafting your own filter modification to match the speaker's natural response in that region if you want to keep just that part of the response mostly unaltered (and this would essentially normalize differences in that dip between speaker locations due to in-room response). And it isn't "enforcing a dip" necessarily. Your speaker might actually have directionality issues in that range that aren't visible in the room due to in-room response, and putting that dip back in might reduce harshness in the vocal range in that case. All of this just gives you the flexibility to tweak things as needed.

If you want a bit more info on MRC and other filter modifications being applied, here's Jeff talking more about it around the 45 minute mark:

Thank you very much.
 

jhaider

Major Contributor
Forum Donor
Joined
Jun 5, 2016
Messages
2,823
Likes
4,522
Below is an example of the difference I measured between Audyssey OFF (red) vs. Audyssey ON (blue) vs. Audyssey FLAT (green).
Were my speakers "designed to have that dip?"

AM-JKLW-EGOCnJ_Msqsu_uXEzTxpAzepwJAyPAO4WWZi8IZ21aHnC4z1cBlHMBooANxEVJ5VOzAkBs8-1byoNN1jMPHzMVcKPZgok1-Cet5bQ72vUNxp1LqHP6W76lFJxqD8oAhzb4MQtepipF_hrjxOvvIOHA=w1024-h598-no

Can you describe your methodology?

If your above measurements are what I assume them to be - single point, but with the microphone kept in the same place - it's hard to do aside from make a few relative comparisons. Still, under my assumption re methodology it's clear you're right that Audyssey's "Reference" target curve introduced a midrange dip beyond what your speakers ordinarily have. I also have some concern this Audyssey target curve generally may be hazardous for your tweeters or your amp. I would use the iOS app (or MultEQ X) to pull the target curve down to follow the speakers' natural response from 3kHz up.

I have my own REW plots that show results consistent with what was explain in that video I referred to. I won't comment on yours as you seem to be too opinionated and not someone who would admit mistakes made, so why bother...

There's a lot of projection there. Still, I see in subsequent posts you've finally decided to accept reality. That doesn't excuse your needlessly hostile tone in your replies to me, but whatever. More understanding and less misinformation is always a good thing.

Yes, at some point you posted some graph salad with no disclosure of methodology or any other outward indication you knew what you were doing in capturing data, processing data, or interpreting the resultant squiggles. You now implicitly admit that, which is good. I won't dwell on that beyond pointing out that tossing a graph salad and then waving in its general direction is not the same thing as carefully and systematically presenting measurements captured employing a sound and disclosed methodology, which were vetted by probably the best Technical Editor in the audio press prior to publication. For reference, since so far as I can tell you still haven't studied my published measurements of Audyssey MultEQ XT32 in Marantz AV7702 or Denon X4100 reviews, or until a little earlier today understood the findings therein regarding the midrange dip in Audyssey's "reference" target curve, here are screenshots of some slides I made using some of those measurements for a training years ago. One note - it looks like a firmware update or something may have corrupted the bass correction as shown in these measurements of the X4100; see review for further analysis:

Denon Audyssey Training Deck Excerpt 1.png


1651098201461.png


1651098169345.png


1651098222282.png

You seem to be one who would repeat misinformation and yet would accuse others of doing that, perhaps that's your strategy??

My "strategy"? What the hell are you alleging? I have no "strategy" here, only an interest in room correction and the continual improvement thereof, and a fair bit of work put in to taking methodology sound measurements of different room correction systems. I do think it inadvertently revealing that you frame our discussion as an adversarial competition requiring "strategy" rather than the way I would frame my approach to any discussion regarding room correction analysis: a cooperative effort to asymptotically approach truth in good faith.

Personally I have zero interest in promoting any specific provider over another; I'd prefer they all be awesome and continually pushing the frontier forward. That means, in part, shining a spotlight on the current limitations of each system. Unfortunately, yet as of nobody has even packaged what I consider the baseline elements in a single room correction system (see below) but a few are farther along that path than others. Those elements are:
  • full incorporation of bass management into the measurement and correction process,
  • multivariable integration of subwoofers and mains that uses relative level, phase/delay, and EQ at the individual-sub level to maximize bass response evenness over the measured area,
  • Easily user-editable target curves of channels that incorporate any bass management in the target curve for each channel and allow for limited correction bandwidth, and
  • ability for the user to lever the correction system for initial setup and confirmatory measurements of all channels (i.e. allow the user to capture spatial average in-room frequency response measurements within the same software)
Some systems may offer more advanced capabilities, and some may also offer side features with potential benefits-in-use but that are not core to system calibration (e.g. loudness compensation). But the above list provides a benchmark that every room correction system should reach or be on a path to reach.
 
Top Bottom