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

[jhaider]

By knowing the standards on general user population and of course testing myself.

Unfortunately I cannot say out loud that I believe many hobbyists exaggerate the importance of smaller details to even themselves. With audible difference question there are blind tests but how can we test does some minor differences in high frequencies…

Here’s where you go off the rails by pretending the elephant in the room is just wallpaper. If you don’t think the average person can hear the difference between room gain accounted for and room gain suppressed, I strongly disagree.

Now, if in sighted listening one is told natural room gain is ”uncorrected” and castrato is “corrected,” I concede a number of people will be swayed more by the words than the sound. However, under blind conditions we know what the actual preference is.

 

[bodhi]

Here’s where you go off the rails by pretending the elephant in the room is just wallpaper. If you don’t think the average person can hear the difference between room gain accounted for and room gain suppressed, I strongly disagree.

Weren't we talking about differences between Audyssey and Dirac? Of course the room gain in lower frequencies in especially small to mid sized rooms can be noticed by a deaf person. But when both Audyssey and Dirac are ran with default settings only? You still think average person, when hearing first Audyssey for half of the movie and then Dirac the other half, would notice how the latter half sounded so much better?

 

[jhaider]

Weren't we talking about differences between Audyssey and Dirac? Of course the room gain in lower frequencies in especially small to mid sized rooms can be noticed by a deaf person. But when both Audyssey and Dirac are ran with default settings only?

Yes. Dirac’s base target curve (especially the new “paddles” version) accounts for room gain and Audyssey’s neuters room gain. Respectfully, this is basic stuff that has been discussed for years - if not for over a decade.

Audyssey can work well in the hands of a dedicated hobbyist who knows how to undo its default decisions as appropriate for her room and configuration . But someone who’s not that dedicated should choose something else.

 

[tjcinnamon]

Yes. Dirac’s base target curve (especially the new “paddles” version) accounts for room gain and Audyssey’s neuters room gain. Respectfully, this is basic stuff that has been discussed for years - if not for over a decade.

Audyssey can work well in the hands of a dedicated hobbyist who knows how to undo its default decisions as appropriate for her room and configuration . But someone who’s not that dedicated should choose something else.

I kinda agree but Audyssey DEQ and Dynamic Volume (which I don’t like) can help consumers fix that “where’s my sub” effect. It can also help with their surrounds. It’s a very plug and play fully integrated system. Becuase of that, despite all of the missteps or unpure decisions, XT32 may have stumbled into an ideal platform for 90% of the users.

Above that, youre correct, a bunch needs to be undone. I find both MQX and Dirac more and more only good for surrounds, gain, and timing. I’d prefer to do my own LCR and Sub on minidsp and REW/Rephase at this point.

I’ll hold steady with my current setup but maybe when ART comes out, I’ll make the switch (if it works). Also, given my towers, that new “all bass” (where LFE goes to mains too) is intriguing (dual 10’s per driver tuned to 23Hz.

 

[jhaider]

I kinda agree but Audyssey DEQ and Dynamic Volume (which I don’t like) can help consumers fix that “where’s my sub” effect. It can also help with their surrounds. It’s a very plug and play fully integrated system. Becuase of that, despite all of the missteps or unpure decisions, XT32 may have stumbled into an ideal platform for 90% of the users.

A correct basic target curve and loudness compensation are not an "either or" situation - they are a "both and" - start with a good target curve and then add loudness compensation to account for lower-than-"ideal" listening levels.

True, units with good loudness compensation and good room correction are in short supply. That's what makes HTP-1 stand out, IMO.

 

[pedrob]

Thanks. Ultimate answer, for those who use manual adjustment without enabling audyssey, is it better to use multeq-x to find the distances and multiply them by 0.875 and insert them into the avr, or use a tape measure and then do the conversion?

I can't thank you enough for this calculation. It made a massive improvement.

I read that Denon uses the speed of sound as 300 m/sec rather than 343m/sec for calculation reasons, and that's how the multiplier was determined.

It seems the speed of sound at 20° C is 344 m/sec. SPEED OF SOUND

I've put my distances in a spreadsheet and multiplied them by 300/344 (0.872 if you prefer) and distances marginally reduced.

It doesn't seem much, but my multi-channelled system is even more in harmony. Almost a sweet spot.

I need a recommendation: For accuracy, I measure my distances with a laser. I've been measuring to the middle of the speaker, but I wonder if I should be measuring to the mid-point between the inner and outer edges of the cone.

BTW, I recall millercarbon describing Home Theater as "crap". Perhaps this is the exact problem he was unable to resolve.

 

[OCA]

I can't thank you enough for this calculation. It made a massive improvement.

I read that Denon uses the speed of sound as 300 m/sec rather than 343m/sec for calculation reasons, and that's how the multiplier was determined.

It seems the speed of sound at 20° C is 344 m/sec. SPEED OF SOUND

I've put my distances in a spreadsheet and multiplied them by 300/344 (0.872 if you prefer) and distances marginally reduced.

It doesn't seem much, but my multi-channelled system is even more in harmony. Almost a sweet spot.

I need a recommendation: For accuracy, I measure my distances with a laser. I've been measuring to the middle of the speaker, but I wonder if I should be measuring to the mid-point between the inner and outer edges of the cone.

BTW, I recall millercarbon describing Home Theater as "crap". Perhaps this is the exact problem he was unable to resolve.

The acoustic center of a speaker is half the cone diameter in front of the cone but that's for the bass (and a bit more complex for ported speakers). Higher frequencies travel faster and arrive first at the LP so I'd go for the tweeter for distance measurement.

PS You can change speed of sound setting in REW to 300m/s and apply all distances measured in REW "as is" to the receiver.

 

[OCA]

A correct basic target curve and loudness compensation are not an "either or" situation - they are a "both and" - start with a good target curve and then add loudness compensation to account for lower-than-"ideal" listening levels.

True, units with good loudness compensation and good room correction are in short supply. That's what makes HTP-1 stand out, IMO.

Unless you listen to your system at 0dBfs which is way louder than anyone can afford to crank up the volume in their homes, it's not the same target curve for all speakers either. DEQ boosts the fronts, surrounds and height speakers differently. The graph below with carefully taken measurements at 75dB illustrates the situation:

Apart from tidying up the lower bass region a bit, it's worse than the uncalibrated sound almost everywhere else. I see Audyssey as an old and buggy calibration system which cannot be further developed by the creators since years due to the license having been sold to Sound United.

But it has one big advantage compared to Dirac. By hacking MultEQ editor app (which was developed by Sound United), the calibration can be fully customized and one can manipulate FIR filters as required in an AVR. For the enthusiast willing to spend time on calibration, the end result will be better than Dirac with which all you can basically adjust is the target curve. Here's what can be achieved with Audyssey (or rather by hacking it):

 

[jhaider]

But it has one big advantage compared to Dirac. By hacking MultEQ editor app (which was developed by Sound United), the calibration can be fully customized and one can manipulate FIR filters as required in an AVR. For the enthusiast willing to spend time on calibration, the end result will be better than Dirac with which all you can basically adjust is the target curve. Here's what can be achieved with Audyssey (or rather by hacking it):

View attachment 287303

I think that analysis misses on two counts.

First and most importantly, a “perfect” looking in room response does not usually (ever?) imply better sound. If you are using good speakers and placing them with care in a non-pathological room, EQing the mids and highs to fit a pre-ordained target curve will only degrade the direct sound and thus the sound quality. Put another way, the speakers should determine the in room midrange and treble response, not you! The highest and best use of software working with listening area data is to mitigate the influence of the room down low and generally match broadband levels. Above, your measurements show loudspeakers with quite a bit more treble energy than one would get from a loudspeaker with flat response under anechoic conditions. However, one presumes you bought them because you liked them.

Second, note the bandwidth of your curve - it’s missing the fundamentals of many notes! Cleaving a channel into two parts is an “original sin” of Audyssey, and in the process kind of defeats the purpose of automated calibration by forcing the user to manually optimize the mains-subwoofers crossover. (There’s also the related issue that Audyssey is still basically a single sub only system - level matching and setting equal delays for two subs isn’t really optimization. You still need a separate box and lots of time and measurements to set up multiple subwoofers.) To be fair, base Dirac does the same thing. The difference is Dirac has better versions (DLBC, ART to come) that treat channels properly as whole entities that happen to be reproduced over multiple transducers, and thereby actually make automated correction automated.

 

[OCA]

EQing the mids and highs to fit a pre-ordained target curve will only degrade the direct sound and thus the sound quality

It does NOT when you apply correct (short enough) windowing to the impulse response, eliminate room reflections in HF and thus correct only for the direct sound. I don't know if it can be seen in the notes under the pictures above but all responses have been applied frequency dependent windowing! I would still never boost the dips in the tweeter crossover range though and I don't do that in my own system but this is not relevant here. The idea was to illustrate Audyssey's odd, channel specific DEQ boosts and deviations from the target curve.

The highest and best use of software working with listening area data is to mitigate the influence of the room down low and generally match broadband levels

That was years ago, digital correction with today's chips and nearly unlimited FIR taps is quite capable of making full band frequency and time domain corrections. It's not straight forward of course and it's not as easy as simply windowing the HF response but it's doable. I agree that you can get 80% there with just simple PEQ below the Schroders's frequency. However, these filters should be minimum phase and we're talking 1024 taps of only linear phase filters with Audyssey.

I fully agree with Audyssey being a single sub system. On top of the problems you have mentioned, it also applies odd boosts to one of the subs for certain channels at certain volume levels. In fact, I ditched my second sub and MiniDSP recently because the extra delays MiniDSP introduced was making it impossible to align subwoofer delay properly and not having to achieve a uniform bass in a large area (my listening area is a two seater sofa), I didn't really get much benefit from the second sub. A fast, precise and perfectly aligned single sub sounds much better overall.

 

[jhaider]

It does NOT when you apply correct (short enough) windowing to the impulse response, eliminate room reflections in HF and thus correct only for the direct sound. I don't know if it can be seen in the notes under the pictures above but all responses have been applied frequency dependent windowing! I would still never boost the dips in the tweeter crossover range though and I don't do that in my own system but this is not relevant here. The idea was to illustrate Audyssey's odd, channel specific DEQ boosts and deviations from the target curve.

I don't agree with the former - in practice your window gets so short with long distances that you have basically no measurement resolution. I think that if you want to EQ the direct field, measure the loudspeaker using standard methods and apply EQ on that basis. Ideally that process should be apart from any automated modal region correction program. Few processors let you do that in practice, though - IIRC just Datasat (though they've been basically dormant since ATI bought them from what I can see; I don't even think they support DLBC), Storm, and Monolith. I think Lyngdorf's PEQ “voicings” are post RoomPerfect but I'm not positive. But most people would need DSP downstream of the processor to do that.

That said, I'm not sure I understand what your graphs are supposed to illustrate. Too much information, too small - even on a 14" retina MBP instead of an iPhone. Regardless, I suspect some like Audyssey's approach to loudness compensation in the surrounds and some don't. Personally, I can't say it bothered me when I used it, though I did prefer Dolby Volume's "modeler" to DynamicEQ at the time.

That was years ago, digital correction with today's chips and nearly unlimited FIR taps is quite capable of making full band frequency and time domain corrections.

“Capable” and “advisable” are sometimes far apart. I’m sure Dr. Toole gets tired of seeing himself incessantly quoted by half-wits on the internet...so I won't add myself to that roster. But I would recommend going back through and reading some of his posts here about room correction, target curves, and magic DSP.

I fully agree with Audyssey being a single sub system. On top of the problems you have mentioned, it also applies odd boosts to one of the subs for certain channels at certain volume levels. In fact, I ditched my second sub and MiniDSP recently because the extra delays MiniDSP introduced was making it impossible to align subwoofer delay properly...

How do you "align" delay? What does that even mean in the context of integrating subwoofers in a small room? Subwoofer delays by themselves have no meaning except in that they can be used as a degree of freedom to achieve a smooth steady state bass response through the crossover and the modal region. Otherwise they're just a good case of the ability of a program to spit out a number without regard to real world utility

A fast, precise and perfectly aligned single sub sounds much better overall.

I’ll leave aside the weird fallacy of “fast” sub…

I can only see two cases where it's plausible that one sub would sound better than two - pathologically bad placement constraints or (more likely) user error. One persistent form of user error is elevating ideology over results.

 

[depechefan]

I purchased a license for Multeq-X for my Denon AVC-X8500HA yesterday. I must say that the difference between the XT32 version and this is night & day. My system sounds fantastic now and I'm hearing things that I never was able to in the past. Can't compare to Dirac as I have no experience with that and it is not an option on my receiver. But I'm very impressed with the new Audyssey version so far. If you have the option and are willing to pay the $200 (more expensive in Denmark...) then I highly recommend it.

 

[dlaloum]

I purchased a license for Multeq-X for my Denon AVC-X8500HA yesterday. I must say that the difference between the XT32 version and this is night & day. My system sounds fantastic now and I'm hearing things that I never was able to in the past. Can't compare to Dirac as I have no experience with that and it is not an option on my receiver. But I'm very impressed with the new Audyssey version so far. If you have the option and are willing to pay the $200 (more expensive in Denmark...) then I highly recommend it.

When using multeq-x, what did you change from the default settings?

 

[depechefan]

When using multeq-x, what did you change from the default settings?

Let me just say that even without changing any settings it sounds way better than anything I have been able to do with the multeq32 app. I usually only correct up to about 500 Hz because it sounds too restrained to me otherwise. But with multeq-x I let it go up to 20 KHz and that sounds awesome
Of course having watched all the videos and playing around I have changed lots of things
In settings I turned on the extra headroom parameter and allowed it to use 12 dB. The 2 settings at the top I have also turned on
I added a house curve by adding a tilt filter (-1 DB sounds good to me. Others prefer around -0.4).
I also added a boost at 60 hz of 1 or 2 DB using a low shelf filter
There is a checkbox that I have checked to override what Audyssey usually does. I can't remember the name. Not in front of my pc...
And I have modified the crossovers
Again: Even defaults sound great. But now it really sings

 

[fieldcar]

When using multeq-x, what did you change from the default settings?

I've found that this is all I need to get what I want.

1. Delete both Theater HF rolloff and Midrange compensation.
2. Add a Tilt and apply a -0.8dB -0.9dB/octave slope (equivalent to a harman 3dB/decade slope), though you can play with this to find your preference.
3. Check "Disable Auto-Leveling" on the subwoofers.

Another cool thing is that If you want to leave the Theater HF Rolloff, just make sure it's set for the reference Type and Set the Tilt type to flat. Then you can switch between both targets on the AVR quickly.

EDIT: Oh yeah, also make sure you take 8 measurement points no further than 6" from the center of your main listening position. I usually cross everything over at 80Hz, and up the subwoofer by a few dB on my HEOS input. You can trim input specific speaker trims by pressing the options? button to the top right of the arrows and going to channel levels or similar.

 

[FrantzM]

I've found that this is all I need to get what I want.

1. Delete both Theater HF rolloff and Midrange compensation.
2. Add a Tilt and apply a -0.8dB/octave slope (equivalent to a harman 3dB/decade slope), though you can play with this to find your preference.
3. Check "Disable Auto-Leveling" on the subwoofers.

Another cool thing is that If you want to leave the Theater HF Rolloff, just make sure it's set for the reference Type and Set the Tilt type to flat. Then you can switch between both targets on the AVR quickly.

EDIT: Oh yeah, also make sure you take 8 measurement points no further than 6" from the center of your main listening position. I usually cross everything over at 80Hz, and up the subwoofer by a few dB on my HEOS input. You can trim input specific speaker trims by pressing the options? button to the top right of the arrows and going to channel levels or similar.

Center frequency for the tilt?

Peace.

 

[FrantzM]

Hi

Quick question. My miniDSP2x4HD died, was my own mistake, long story.
I want to use the 2 subs in the system. Will MultEQ-X (Windows app) be able to calibrate the 2 subwoofers properly? Taking care of the different delays and EQ parameters for each subwoofer, a la MSO? I have one sub in the front and the other in the back.


Peace.

 

[fieldcar]

Center frequency for the tilt?

Peace.

I leave it at 1KHz, The only time I think you need to worry about leaving it at 1KHz is if the compensation EQ maxxes out on the Filter settings graphs. If that's the case, just go to the settings at the lower left > features tab at the top > Enable EQ Headroom Expansion.

 

[depechefan]

This is what I have. My understanding is that there is a correlation between the "Enable HQ Headroom Extension" setting and "Disable Auto Leveling". I have this disabled for all my speakers. I think it sounds good and that is the most important factor for me - and hopefully everybody else as well

 

[FrantzM]

Hi

Quick question. My miniDSP2x4HD died, was my own mistake, long story.
I want to use the 2 subs in the system. Will MultEQ-X (Windows app) be able to calibrate the 2 subwoofers properly? Taking care of the different delays and EQ parameters for each subwoofer, a la MSO? I have one sub in the front and the other in the back.


Peace.

Bump!
?