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

[tjcinnamon]

You can do better way than that (with MultiEQX) provided your speakers (and MLP) can do it. Or is this smiley face frequency response how you prefer it?

I get pretty much Harman preference curve (with ca 4 dB low frequency boost) with my setup.

I do like it bright. Also, DEQ probably bumped the tail up like that. My kids were sleeping so I measured at -30dB where I listen at -20dB.

I’ve got another measurement that’s wicked flat on a different present. This one sounds better than that one. Could be the phase, could be the fact that they overlap so smoothly. But this one sounds better. That hump around 1000 presents itself really easily. So that’s a room thing.

Some folk don’t EQ above 300. I’m not in that camp.

 

[tjcinnamon]

Can i find you on social or somewhere to chat a bit about multeq-x? If yes here is my fb page if you can sent me a message i would really appreciate it.

If you cant/wont go into social, can you please answer any of my previous questions, seeing that you are using multeq-x with rew?
Just for clarification this is the order of the question's in my previous post:

1. Can we completely disable audyssey corrections? (If yes, do we still have deq and dyn volume?)
2. Is the processing of the rew filters done in a "pure" way?
(Practically Meaning: do you get the predicted result of the filters when designing them in rew when you remeasure with rew after you apply them in multeqx?)
(Theoretically Meaning: do you know if the filters are processed as is or does multeq-x converts them and tries to approximate them?)
3. Do you happen to know the differences/implications of the multeq-x use in older receivers with the multeq xt version instead of the xt32?

I appreciate any knowledge IF you happen to have in any of the above questions!

I’m not on social but will happily answer. For 1: yes and yes. I’m a big fan of DEQ and like my own curve.
2: yes, I get the predicted results within reason. I do measure my source measurements with DEQ engaged and at my standard listening level. Keepin mind that I’m using an average of measurements for the source and replicating that to “check my work” is cumbersome (and I don’t do it).

3: I’m unsure but I’m sure there is documentation. I’m not even sure it’s compatible. MQX might require more horsepower for the FIR filters it creates

 

[JohnAps]

I’m not on social but will happily answer. For 1: yes and yes. I’m a big fan of DEQ and like my own curve.
2: yes, I get the predicted results within reason. I do measure my source measurements with DEQ engaged and at my standard listening level. Keepin mind that I’m using an average of measurements for the source and replicating that to “check my work” is cumbersome (and I don’t do it).

3: I’m unsure but I’m sure there is documentation. I’m not even sure it’s compatible. MQX might require more horsepower for the FIR filters it creates

I have searched everywhere, and i have even deployed bing chat and chatgpt4 browsing plugin for this and none of us can find anything as per documentation on any of the above matters.
What i want to establish basically is if the imported filters do work as FIR filters "as they should". Or if multeq-x is using the filters to create a "target curve" as the mobile app is doing and then applies that correction as best it can...
Because if its the latter then i wont get any improvement on my speaker's response compared to the mobile app + rew custom curves i am currently doing...

 

[JohnAps]

I’m not even sure it’s compatible. MQX might require more horsepower for the FIR filters it creates

Supposedly it works in any receiver that the mobile app works! Thats what the documentation says at least and i have verified that multieq-x does see my denon

 

[Dobbyisfree]

3. Do you happen to know the differences/implications of the multeq-x use in older receivers with the multeq xt version instead of the xt32?

It's compatible with loads of models with only XT and, amazingly, models with only MultEQ too (not XT).

A guy on AVForums shared the results from MultEQ-X from his Marantz Cinema 70 that only has MultEQ.

I have the same EQ on my 2009 Denon AVR1910 (used in a spare bedroom) . After seeing his (visual) results on the Marantz I can see why the results (audibly) on the old Denon were so poor!

 

[HarmonicTHD]

I do like it bright. Also, DEQ probably bumped the tail up like that. My kids were sleeping so I measured at -30dB where I listen at -20dB.

I’ve got another measurement that’s wicked flat on a different present. This one sounds better than that one. Could be the phase, could be the fact that they overlap so smoothly. But this one sounds better. That hump around 1000 presents itself really easily. So that’s a room thing.

Some folk don’t EQ above 300. I’m not in that camp.

Ah DEQ explains it. That’s fine. Without that information people might get the wrong impression that this would be the performance you get with Audyssey/ MultiEWx.

 

[tjcinnamon]

Ah DEQ explains it. That’s fine. Without that information people might get the wrong impression that this would be the performance you get with Audyssey/ MultiEWx.

That’s a very good point! I’ll post my super flat one. To make up for it. It’s I botched the sub on that one but really wasn’t focusing on that at the time. I run my 5 subs in a minidsp and have MQX completely skip them.

It sounds fantastic BTW! I’m a huge fan of DEQ. Once I post my sub measurement, people will think I’m insane. I run 3 DIY bridges with Crown XLS amps about 9inches from MLP, the SPL is crazy because of the close measurement but the decay time is low (relatively speaking) because the SPL is lower in the room.

I wish I had a 4800 (instead of a 4700) so I could A/B Dirac and MQX. I did have a Dirac NAD unit for 3 years and ran countless calibrations with that. And I like MQX better for now.

 

[tjcinnamon]

I have searched everywhere, and i have even deployed bing chat and chatgpt4 browsing plugin for this and none of us can find anything as per documentation on any of the above matters.
What i want to establish basically is if the imported filters do work as FIR filters "as they should". Or if multeq-x is using the filters to create a "target curve" as the mobile app is doing and then applies that correction as best it can...
Because if its the latter then i wont get any improvement on my speaker's response compared to the mobile app + rew custom curves i am currently doing...

Yes, they work as they should. There’s a high probability there would be an improvement. it corrects an issue with the delay. The internal iterations of Audyssey use the incorrect formula for the speed of sound. MQX corrects that.

I have no experience with the mobile app.

 

[JohnAps]

it corrects an issue with the delay. The internal iterations of Audyssey use the incorrect formula for the speed of sound. MQX corrects that.

The issue on delay can be solved by multipling the distances in meters with 0.875 manually if you dont have MQX, so i dont have that problem...

 

[JohnAps]

Yes, they work as they should. There’s a high probability there would be an improvement.

For example even if you make many filters and some of them very steep like 20Q, have you verified that they are applied as you imported them?

 

[JohnAps]

It's compatible with loads of models with only XT and, amazingly, models with only MultEQ too (not XT).

A guy on AVForums shared the results from MultEQ-X from his Marantz Cinema 70 that only has MultEQ.

I have the same EQ on my 2009 Denon AVR1910 (used in a spare bedroom) . After seeing his (visual) results on the Marantz I can see why the results (audibly) on the old Denon were so poor!

Do you have more specific examples relatively to the application of the filters in older models?

 

[tjcinnamon]

Ah DEQ explains it. That’s fine. Without that information people might get the wrong impression that this would be the performance you get with Audyssey/ MultiEWx.

Disregard the poor left measurement for the sub.

 

[HarmonicTHD]

Disregard the poor left measurement for the sub.

Better. But 20dB !!! Bass boost. Wow.

Well if you like it. Enjoy.

 

[chych7]

Hello!
Seeing some of @chych7 &
@tjcinnamon replies of the forums i have established that we can disable audyssey corrections for our speakers and work by importing filters and/or target curves from rew, to our speakers for correcting them...
I have a denon avr-2400h with audyssey multeq xt and i wanted to ask of you if the the multeq-x is indeed working with the the rew filters as it should. Meaning do we get the predicted result if we measure again with rew after the import?
Are the filters we upload used as they are or does it convert them to a curve and computes its own filtering or something (probably meaning we dont get exact results from rew - multieqx integration).
Lastly do you happen to know if in older receivers like mine (that have the plain old multeq xt), do these or other important functionalities, like the filter import etc, work in a worse manner?

I am looking to buy the program, but there arent many verifiable information about this.
I was using rew + mobile app with some perfect cal files "hacks" to achieve custom correction while disabling audyssey corrections, and i want to find out if i can do the same but better with multeqx, which i will if multeqx can indeed and truly work with filters and it isnt some gimmicky conversion of the filters which results in the same behavior as the mobile app target curve function...

I sincerely thank you for any conclusive information on this, if you are actively using rew and multeq-x for manual calibration as you will help me have a clear image of the program before i buy.

In my latest calibration with the headroom expansion feature turned on, I found the default MQX calibration worked really well, so I didn't bother REW filter import. But generally I did see the REW filter import to work very well; what was created in REW was very close to the final result after measurement. MQX imports the REW EQ and creates its own correction, from what I can gather.

XT should still have enough capability to correct mids-highs, so it could help for correcting speaker response and matching speakers to each other, which can help with imaging. XT is less capable than XT32 for bass correction. Honestly not sure if it's worth the $200 for it without XT32. There are ways to get similar performance with the app (

)

 

[JohnAps]

In my latest calibration with the headroom expansion feature turned on, I found the default MQX calibration worked really well, so I didn't bother REW filter import. But generally I did see the REW filter import to work very well; what was created in REW was very close to the final result after measurement. MQX imports the REW EQ and creates its own correction, from what I can gather.

XT should still have enough capability to correct mids-highs, so it could help for correcting speaker response and matching speakers to each other, which can help with imaging. XT is less capable than XT32 for bass correction. Honestly not sure if it's worth the $200 for it without XT32. There are ways to get similar performance with the app (

)

That is what i am trying to figure out @chych7 !
Actually the video you uploaded is OCA's which i showed him the perfect ady file method from another member on one of the forums, which in turn inspired his video!

I have done this calibration method in the video (and i can say i have done it with more precision than it with some variations in the method) but cause of the limited amount of points we can make to export our inverted target curves, and/or the limits of the receiver on how its using this target curve, there isnt a very good correction, i would say, on my receiver i get about 50% better responses this way...

There are a few features that make multeq-x appealing to me, but the most important one is if the rew imported filters will work better than what we can achieve through the app and the methods in the above video or not...

(Some other members as well have said that the auto method of multeq-x is better than stock, but for now i am not interested in that so much.)

Thankfully i am now talking with a guy familiar with my situation so i hope you can give me some insights that will steer me into the right direction

 

[chych7]

That is what i am trying to figure out @chych7 !
Actually the video you uploaded is OCA's which i showed him the perfect ady file method from another member on one of the forums, which in turn inspired his video!

I have done this calibration method in the video (and i can say i have done it with more precision than it with some variations in the method) but cause of the limited amount of points we can make to export our inverted target curves, and/or the limits of the receiver on how its using this target curve, there isnt a very good correction, i would say, on my receiver i get about 50% better responses this way...

There are a few features that make multeq-x appealing to me, but the most important one is if the rew imported filters will work better than what we can achieve through the app and the methods in the above video or not...

(Some other members as well have said that the auto method of multeq-x is better than stock, but for now i am not interested in that so much.)

Thankfully i am now talking with a guy familiar with my situation so i hope you can give me some insights that will steer me into the right direction

I haven't tried OCA's method so I don't have the full experience on that. Back when I tried doing manual target curve adjustments with the app and ratbudyssey, I wasn't able to get great results. MQX gave me good results out of the box and is fairly straightforward to use. The recent version of MQX allow for headroom expansion and higher cuts (see youtube video). I think this would be a real enhancement to Audyssey's ability to EQ, which cannot be achieved with the phone app method. But, I'm not sure how this applies to XT, my experience is from XT32.

From REW sweeps, how well matched are your speakers with the calibration you've done so far? There may not be much more to gain with MQX.

 

[JohnAps]

I haven't tried OCA's method so I don't have the full experience on that. Back when I tried doing manual target curve adjustments with the app and ratbudyssey, I wasn't able to get great results. MQX gave me good results out of the box and is fairly straightforward to use. The recent version of MQX allow for headroom expansion and higher cuts (see youtube video). I think this would be a real enhancement to Audyssey's ability to EQ, which cannot be achieved with the phone app method. But, I'm not sure how this applies to XT, my experience is from XT32.

From REW sweeps, how well matched are your speakers with the calibration you've done so far? There may not be much more to gain with MQX.

The matching correction isnt great... as i said even viewing before and after with phychoacoustic smoothing you will see a 50% improvement of the whole curve closeR to my target curve, but there are 2-5db dips and peaks almost everywhere still, making my response pretty far from smooth or straight...

But i fear to give 200 dollars and not achieve a better correction, thats i why i am trying to find out if the application of filters through MultEQ-X is more precise than what we can do with the custom method and the mobile app, thus better.

I tend to believe based on the ui and how they advertise it, that we would get better correction from the software but i am not sure as i am not fully knowledgeable on how everything works...

 

[JohnAps]

Also do you happen to know if as i concluded looking at the ui, can increase the crossover of speakers higher so they send the low frequencies to sub, BUT still being able to modify the cutoff point to a different point that the set crossover, meaning that if we want we can still make a speaker play full range, but also sent the low frequencies to the sub if we have set the crossover higher? This is something that interestes me as well and it seems sure it can happen with MQX but i would like some verification!?

 

[chych7]

The matching correction isnt great... as i said even viewing before and after with phychoacoustic smoothing you will see a 50% improvement of the whole curve closeR to my target curve, but there are 2-5db dips and peaks almost everywhere still, making my response pretty far from smooth or straight...

But i fear to give 200 dollars and not achieve a better correction, thats i why i am trying to find out if the application of filters through MultEQ-X is more precise than what we can do with the custom method and the mobile app, thus better.

I tend to believe based on the ui and how they advertise it, that we would get better correction from the software but i am not sure as i am not fully knowledgeable on how everything works...

Yeah I can't give a definitive answer on this. I'll take some measurements of my calibration to show you what I got.

There may be some Microsoft refund policy to try and get a refund if it doesn't work out:

How to request Microsoft Store refunds on Windows 10 and 11

Get your money back.

www.windowscentral.com

Also do you happen to know if as i concluded looking at the ui, can increase the crossover of speakers higher so they send the low frequencies to sub, BUT still being able to modify the cutoff point to a different point that the set crossover, meaning that if we want we can still make a speaker play full range, but also sent the low frequencies to the sub if we have set the crossover higher? This is something that interestes me as well and it seems sure it can happen with MQX but i would like some verification!?

No, the crossover would override the cutoff in this case. You could set speakers to Large, turn on LFE+main, which would send bass to both speakers and subwoofer. Then you can modify the cutoff point on the speaker and play with the EQ to try and integrate it well. I haven't tried this, sounds challenging.

 

[chych7]

These were my speaker results with the latest MQX version, using the ACM-1X mic. My target curve was just -0.8/dec slope (similar to what Dirac uses).

DynamicEQ was on in these measurements, which creates an uplift in the bass and treble, so it's not quite matching the downward slope target. I subtracted 3 dB from the surround channel levels to compensate DEQ surround boost.

The dip/peak around 1-2 kHz is due to my seat's high headrest. It cannot be EQ'd out, I've tried.... and it changes as the seat recline/headrest is adjusted.

But in any case, I think the 2-5 dB peaks/dips are typical for in room response. Getting a ruler flat response is probably not realistic. Also importantly, Audyssey measures multiple locations and averages. When doing REW verification, it's going to be difficult to verify at the exact same measurement points. A single point measurement isn't really going to tell you about the response. Heck, just having your body sitting in the seat is going to change the response a fair bit.

I am using a miniDSP with MultSubOptimizer for bass, not showing that as it's not part of the MQX calibration.


All base channels

Just LR:

LR in pure direct mode/Audyssey off: