Big news coming from Sound United in 2023!

[Jon AA]

I chose my words carefully. “The general AVR RC products” are not the right tool to fine tune a speaker’s anechoic full range response.

If you use said products the way the manual says to use them, you will make a good speaker sound worse if you allow full range correction.

That's still too absolutist for me to agree with. This thread isn't about "general AVR RC products," it's about a product line that will offer Audyssey and Dirac. Both of these allow and encourage custom target curves. All the user needs to do is select a target curve that improves the response of the speaker.

Without anechoic data, it becomes a semantics question of where you place the bar for a "good speaker." Since I'll always use EQ, the natural on axis frequency response is down the list on items of importance for me. When I see a speaker that has well controlled, smooth directivity, I'll say that's a "good speaker" even if it has some serious errors in its on axis frequency response. We've seen quite a few speakers test that way here.

For a speaker like that, it's actually quite likely you can make it sound better with just a generic Harman-like room curve. When you correct it to a smooth, sloping line you will be making the direct sound more linear if the speaker has good directivity. Adjust the tilt with some listening tests, and the sound very likely has improved. It won't be perfect without more effort, but the blanket statement that you will make it worse is just not true. A speaker with directivity issues? Yes, it's less likely you'll be successful without anechoic data, but I'd also say that's not really a "good speaker" either.

Since I'm a perfectionist, I like to use anechoic data to help find the room curve that equates to flat, smooth, direct sound a bit more precisely. And when you do that, room EQ is a fine tool to "fix" your speakers as well as they can be fixed.

 

[Jon AA]

How do you work around the problem that if you EQ only bass, and then boost bass in the target curve, it doesn't actually boost the bass (or doesn't boost it by much) as you're trying to boost the "entire range" (out of the EQ-enabled range, or at least a large portion of it)?

That brings up a good point I should have mentioned--you always want to measure afterward to verify things line up correctly. With this flaw in the way Audyssey sets the levels and doesn't do the speaker and sub together to make sure they match at the crossover, you always want to be on the lookout for it so you can adjust if needed. Usually, if done as I suggest things do, but not always. Look for a response that follows the target curve through the crossover like so and adjust if necessary (an example of my left front speaker, crossover 60 Hz):

Basically, if you EQ to 300Hz and boost bass up to 200Hz by 10db, it will actually not perform what you intended it to.

There could be many reasons for this. First, 10 dB is a lot. I believe the max boost Audyssey will do is 9 dB, so you'd need a fair amount of natural bass reinforcement to have a hope of getting that much. Secondly, the most likely reason it doesn't reach your intended bass boost is the LF Cutoff setting. If it's applying a second order high pass, etc, the speaker may never get all the way to the max boost you need. You may need to lower the frequency of it, change it to 4th order or override it completely to get that much boost.

My understanding is that Auydyssey sets the levels based upon an average from 500-2K Hz for the main channels, so the bass boost below that shouldn't affect the level. However, as you may have noticed, changing the EQ limit above that frequency can. It's common for the speaker to have pretty large dips/peaks all the way past 500 Hz and as you move that curtain you are adding or subtracting them from the average which can affect the overall level. But it's usually within +/- 1 dB or so, so that wouldn't explain 10 dB. Keep in mind the EQ limit is not a brick wall--it continues to EQ an octave above where you set it, gradually reducing the magnitude of the filter adjustments, so moving around that EQ limit anywhere between 250-2K Hz can change the level a bit. But yes, EQing full range so you have a smooth response through that range generally takes care of the issue.

Out of curiosity I changed one of my old calibrations to have 10 dB of boost, here is the correction it shows with a limit of 300 and full range:

As you can see, it never gets to 10 dB, but does get close. Changing the EQ limit does change the level, but it's a small amount.

Do other EQ systems (Dirac, Arc) have similar limitations or do they work differently?

I haven't followed Arc for years, but at the time they were having all sorts of problems around issues like this. But that was years ago, I'd assume they have it fixed by now. With DLBC it certainly shouldn't be an issue with Dirac as it does the speaker/sub together for the channel which is a real advantage. I don't think the base level version does, but I haven't kept current on that either.

I checked this by using a friend's MultEQ-X software with his measurements to try help him improve his sound.

If you'd like to send me the file I'd be happy to take a look, you've got me curious.

 

[Jon AA]

Besides, I am afraid that this enhanced app would not solve the bass issue for me.

MultEQ-X could certainly fix it.

You should hear two deep "thumps" at the beginning, the second deeper than the first (I did not conduct a spectral analysis to see how deep they go, but I would guess 20-25 Hz).

OK, this is a clue as to what may be going on. What exact speakers do you have? 20-25 Hz is really low for a tower speaker. I'm guessing that Audyssey is detecting a drop in output at a higher frequency from at least one of them. If it does, it will apply a high pass filter to both of them to protect them from damage. This cannot be fixed with the App, but can with MultEQ-X. If you are confident your listening level won't damage the speakers, with MultEQ-X you can lower the cutoff, change the high pass all the way up to a 6th order, or eliminate it all together. If your speakers are capable of playing it, MultEQ-X will allow it.

 

[lc6]

How are the speaker measurements before and after Audyssey @lc6? You mentioned that DEQ operation was clear on REW measurements.

I was referring to the measurements referenced in my post #718 and @-Matt- 's post #722 in this thread.

 

[Newman]

That's still too absolutist for me to agree with. This thread isn't about "general AVR RC products," it's about a product line that will offer Audyssey and Dirac.

So you like to take my comments out of context and put them into your context instead of the one I stipulated? Don’t be absurd. No wonder this thread is endless.

Take another look. I was responding —23 pages ago!— to a specific comment that wished for some more point-and-shoot advice for “Joe Sixpack”, instead of the entangled techniques that characterise this thread. I’m allowed to do that without you coming along hundreds of posts later and dragging it all completely out of context, and now playing thread cop to boot. Give it a rest.

 

[lc6]

MultEQ-X could certainly fix it.


OK, this is a clue as to what may be going on. What exact speakers do you have? 20-25 Hz is really low for a tower speaker. I'm guessing that Audyssey is detecting a drop in output at a higher frequency from at least one of them. If it does, it will apply a high pass filter to both of them to protect them from damage. This cannot be fixed with the App, but can with MultEQ-X. If you are confident your listening level won't damage the speakers, with MultEQ-X you can lower the cutoff, change the high pass all the way up to a 6th order, or eliminate it all together. If your speakers are capable of playing it, MultEQ-X will allow it.

Do you rather mean "I'm guessing that Audyssey is detecting a drop in output at a higher lower frequency from at least one of them"?
It is very interesting to learn about this Audyssey "nanny" feature which, just like the "mid-range correction," does not appear to be described anywhere in the 3800H owner's manual.

I am purposely not disclosing the makes/models of my speakers or headphones because I do not want this discussion to veer off into an evaluation/criticism of them instead of the 3800H. Some members appear to be really determined to prove that the D/M gear is just perfect (especially if they own some models of it), so the fault must be with the user or room or speakers, or all of the above.

Again, I very much appreciate your suggestion to use MultEQ-X, but from my perspective that would add several hundred dollars to the price of this AVR with no guarantee of a satisfactory outcome (your confidence notwithstanding). Also, assuming that the claims that Dirac Live is so much better than Audyssey (presumably, even the MultEQ-X version) are true, then next spring I would have to pony up yet another several hundred dollars for a new room correction software and throw away the MultEQ-X. So, I'd rather apply the cost of MultEQ-X + its calibrated mic toward a future upgrade of my height speakers or a purchase of rear surrounds (which I currently do not have).

Now, a quick recap of my user experience with the HEOS app last night: It does work with Amazon Music HD (AMHD), but has a limited function. Unlike with the AMHD apps on the Fire Stick or TV, only the format of the source material (e.g. 24 bits / 96 kHz) is shown, but not the actual playback parameters. So there is no way to tell if the AVR actually plays hi-res streams and at what sampling rate. There is a cryptic Quality Low/High setting, but no explanation what it actually does (e.g. Low = lossy Opus vs. High = lossless FLAC).
One of the key features in missing: an ability to look up the entire album from which the track being played comes, or the artist profile to access their albums and select a different one. This feature is supported on the Fire Stick and TV AMHD apps.
Also, it is unclear to me how this HEOS app actually works in conjunction with the AVR. Does the music stream go directly from the Internet to the AVR (with the app only acting as a graphical user interface and AVR controller), or does the stream first go to the app which then sends it over the LAN to the AVR? I was unable to determine that on the device running the app.

 

[Newman]

Without anechoic data, it becomes a semantics question of where you place the bar for a "good speaker."

?? A good speaker is one for which you do have anechoic data on and off axis and you know it meets the parameters that have been experimentally determined to be pretty much universally preferred.

Since I'll always use EQ, the natural on axis frequency response is down the list on items of importance for me. When I see a speaker that has well controlled, smooth directivity, I'll say that's a "good speaker" even if it has some serious errors in its on axis frequency response. We've seen quite a few speakers test that way here.

I’ll tell you what, they are not good speakers in my book, which BTW is consistent with the literature. So I don’t think your saying it is good matters, when it simply isn’t.

For a speaker like that,

…ie a not-good speaker, so not the sort of speaker I was referring to…

it's actually quite likely you can make it sound better with just a generic Harman-like room curve. When you correct it to a smooth, sloping line you will be making the direct sound more linear if the speaker has good directivity. Adjust the tilt with some listening tests, and the sound very likely has improved. It won't be perfect without more effort, but the blanket statement that you will make it worse is just not true.

That’s only because you are starting with a not-good speaker, by definition, which is the opposite of what I was referring to.

A speaker with directivity issues? Yes, it's less likely you'll be successful without anechoic data, but I'd also say that's not really a "good speaker" either.

Well duh!

Since I'm a perfectionist,

LOL, if you use the method you describe above then you are a perfectionist at taking a compromise approach to extremes.

I like to use anechoic data to help find the room curve that equates to flat, smooth, direct sound a bit more precisely. And when you do that, room EQ is a fine tool to "fix" your speakers as well as they can be fixed.

Absolutely not. A perfectionist takes a speaker with good on and off axis behaviour outside the bass, and avoids room correction outside the bass, because it’s only going to make things worse. A ‘compromiser’, OTOH, starts with compromised speakers that are not good on-axis, then instead of taking direct anechoic measurements on-axis and applying the correct PEQ to smooth it in the areas it should be smoothed, which is what the perfectionist would do with not-good speakers, instead, the compromiser tries to make proxy backwards adjustments using ungated measurements and guessing on what ungated target curve might, hopefully, with luck, move things in the right direction, by proxy assumptions.

I’ll never endorse that!

 

[Newman]

Do you rather mean "I'm guessing that Audyssey is detecting a drop in output at a higher lower frequency from at least one of them"?
It is very interesting to learn about this Audyssey "nanny" feature which, just like the "mid-range correction," does not appear to be described anywhere in the 3800H owner's manual.

I think he is simply describing the basic hygiene check that all RC products do, in guessing the LF limit of the speakers, and not boosting bass below that frequency. Otherwise speakers 20dB down at (say) 40 Hz will get 20 dB boost at 40 Hz, and that will destroy a lot of them. That’s not a ‘nanny feature’, it’s essential.

Does your Yamaha RC also do this? Most certainly.

Is your Denon guessing a higher LF limit than your Yamaha? Possibly.

Is your Denon guessing a higher LF limit for one of your front speakers than the other, and applying that to both? No idea.

 

[Chromatischism]

Do you rather mean "I'm guessing that Audyssey is detecting a drop in output at a higher lower frequency from at least one of them"?
It is very interesting to learn about this Audyssey "nanny" feature which, just like the "mid-range correction," does not appear to be described anywhere in the 3800H owner's manual

Audyssey detects the -3 dB point of your speakers and does not EQ below that. It is important because you don't want to overdrive speakers below their natural rolloff by treating it the same as any other "dip".

I am purposely not disclosing the makes/models of my speakers or headphones because I do not want this discussion to veer off into an evaluation/criticism of them instead of the 3800H. Some members appear to be really determined to prove that the D/M gear is just perfect (especially if they own some models of it), so the fault must be with the user or room or speakers, or all of the above.

The reason is because if you really have full range speakers producing 20 Hz (a story which conflicts with your Audyssey results, which we haven't gotten to the bottom of yet), they need hundreds of watts, which an AVR isn't going to provide, especially if the impedance drops to 4 ohms or less. This is a big reason (but not the only) why we use subs.

Also, assuming that the claims that Dirac Live is so much better than Audyssey (presumably, even the MultEQ-X version) are true,

It's better in some ways, worse in others. "So much better" is not what I would call it.

Now, a quick recap of my user experience with the HEOS app last night: It does work with Amazon Music HD (AMHD), but has a limited function. Unlike with the AMHD apps on the Fire Stick or TV, only the format of the source material (e.g. 24 bits / 96 kHz) is shown, but not the actual playback parameters. So there is no way to tell if the AVR actually plays hi-res streams and at what sampling rate. There is a cryptic Quality Low/High setting, but no explanation what it actually does (e.g. Low = lossy Opus vs. High = lossless FLAC).
One of the key features in missing: an ability to look up the entire album from which the track being played comes, or the artist profile to access their albums and select a different one. This feature is supported on the Fire Stick and TV AMHD apps.
Also, it is unclear to me how this HEOS app actually works in conjunction with the AVR. Does the music stream go directly from the Internet to the AVR (with the app only acting as a graphical user interface and AVR controller), or does the stream first go to the app which then sends it over the LAN to the AVR? I was unable to determine that on the device running the app.

The app is only a controller. The AVR is pulling in the stream itself.

 

[Chromatischism]

Absolutely not. A perfectionist takes a speaker with good on and off axis behaviour outside the bass, and avoids room correction outside the bass, because it’s only going to make things worse.

Actually, I only agree with this half way.

For those who have a well-set up system, I agree. Symmetrical, stereo triangle, equal walls, no coffee table, etc.

But for many homes that don't have that, room correction can correctly reduce the total sound power of a frequency range which is coming from a strong reflection or an uneven room. It is very noticeable at my computer, where I reduce the desk reflection and dramatically improve the sound.

So on this topic, I don't speak in absolutes, as I used to

 

[lc6]

I think he is simply describing the basic hygiene check that all RC products do, in guessing the LF limit of the speakers, and not boosting bass below that frequency. Otherwise speakers 20dB down at (say) 40 Hz will get 20 dB boost at 40 Hz, and that will destroy a lot of them. That’s not a ‘nanny feature’, it’s essential.

Does your Yamaha RC also do this? Most certainly.

Is your Denon guessing a higher LF limit than your Yamaha? Possibly.

Is your Denon guessing a higher LF limit for one of your front speakers than the other, and applying that to both? No idea.

My old Yamaha AVR does not have RC or adjustable EQ functions; it only has a hardwired and non-adjustable loudness compensation, just like most conventional amps do. So it is rather improbable that it has any protective HPFs in the bass region. It has been driving the same front speakers for ~23 years without problems (yes, they do not make the audio equipment as they used to ). No damage; the speakers still sound as fine at the lowest end as they originally did; they are rated up to 250 W, which I am very far from ever reaching.

The proper approach in an RC-equipped AVR would be to simply not correct (boost) the FR below a certain "nanny" frequency, but still pass through the original-level signal to the speaker instead of filtering it out as the 3800H is apparently doing. Now, if the 3800H actually does the former and the lowest end sounds too weak, then this would point to the amp not having a sufficient current (yes, I know @-Matt- will hate me for saying this again ).

 

[peng]

Oh God please no! A flat listening position response only sounds good to people with massive hearing loss. It's also very dangerous to tweeters. I've measured more than 6dB HF boost.



Ditching the dip is the only good thing "flat" does.

View attachment 235072

I looked at my rew curve inventory and they show "Flat" curve did not disable/or "ditch" MRC, will plot some fresh curves to check again. I also asked "Ask Audyssey" and will post their response. Since you are the expert, I would assume you are right, until and unless I have more information/evidence to share.

 

[peng]

Oh God please no! A flat listening position response only sounds good to people with massive hearing loss. It's also very dangerous to tweeters. I've measured more than 6dB HF boost.



Ditching the dip is the only good thing "flat" does.

View attachment 235072

Okay, managed to find some time so I plotted some REW graphs, my results show "Flat" does not disabled the MRC dip unless I disabled it with the Editor App:

You can see that in my graph below, at around 2,000 Hz, reference and flat target curve have virtually identical response.

I still have the mic and laptop hooked up and will leave it there for another hour or so in case someone wants to see other plots of X vs Y.

 

[kryptonite]

May I recommend that this thread remain specific to news, specs and info about the new Denon 2022 series of AVRs and that Audyssey related discussions (not specific to the new Denons) be moved out to another thread? I think both deserve equal merit and some insights into Audyseey might lose their importance in this thread.

 

[-Matt-]

May I recommend that this thread remain specific to news, specs and info about the new Denon 2022 series of AVRs and that Audyssey related discussions (not specific to the new Denons) be moved out to another thread? I think both deserve equal merit and some insights into Audyseey might lose their importance in this thread.

I agree with the sentiment but I'd already written the following (it was to be my last attempt to help in any case).

... then this would point to the amp not having a sufficient current (yes, I know @-Matt- will hate me for saying this again ).

No hate here!

What was the power rating of your previous Yamaha? Which model was it?

No damage; the speakers still sound as fine at the lowest end as they originally did; they are rated up to 250 W, which I am very far from ever reaching.

I think this is the first time you mentioned the power handling capability of your speakers (you haven't given us much information to go on). You did say, I think, that you don't listen too loud, but more recently mention 85dB during workouts which is rather loud. Power requirements also depend on the size of the room (again, not much info).

A major differentiator between Denon AVRs: 3x00, 4X00, 6X00, 8X00 etc is the power per channel. It is possible that you would have been better served by one of the higher power ones but we can only guess unless you give details of the speakers and room.

Even if that turns out to be the case, I still don't think it gives a valid basis for a review saying that the 3800 lacks bass. It would rather be a case of selecting an inapproriate amplifier for large speakers. Many users with smaller speakers (the 3800 is intended for modest setups) would have no such problem.


If you are still interested in trying to improve the bass:
...It sounds as though you are still insisting on using LR bypass which is actively preventing Audyssey from being able to improve the response of your mains. I'd use the Audyssey reference setting instead.

Please also confirm that settings which are intended to reduce bass (Low Frequency Containment) or reduce dynamic range (Dynamic Volume) are turned off. Also Eco mode off.

Also, we may have a slight terminology mismatch! You complain that there is not enough bass but also dislike the low thumping sounds. Most of the measures suggested here have been aimed to increase sub-bass (which most of us want). It seems that this may not be to your taste and that perhaps what you want more of is more mid-bass? Please clarify. It would be easy to see such a failing if you could plot a frequency response graph (cheapest way of getting this is with the $20 app).

 

[Jon AA]

I’m allowed to do that without you coming along hundreds of posts later and dragging it all completely out of context, and now playing thread cop to boot. Give it a rest.

Wah. If you want to make absolute statements in a way Toole typically doesn't in his name, you had better be pretty sure about them and be willing to back them up.

A perfectionist takes a speaker with good on and off axis behaviour outside the bass, and avoids room correction outside the bass, because it’s only going to make things worse.

And you just repeated it. It's provably wrong. When you have anechoic data, very, very, few speakers are so good they can't be improved.

…ie a not-good speaker, so not the sort of speaker I was referring to…

Like I said, semantics. For me, the most expensive speakers Revel makes aren't flat enough (or Kevin Voecks), none of the JBL synthesis line is (for me or the Engineers at JBL), etc. If a perfectly flat LW is a requirement, the list of "good speakers" in existence is quite short indeed.

FR can be fixed with EQ, directivity cannot. And yes, virtually every passive speaker I've seen measured could use "some fixing."

LOL, if you use the method you describe above then you are a perfectionist at taking a compromise approach to extremes.

False.

Absolutely not. A perfectionist takes a speaker with good on and off axis behaviour outside the bass, and avoids room correction outside the bass, because it’s only going to make things worse. A ‘compromiser’, otoh, starts with compromised speakers that are not good on-axis, then instead of taking direct anechoic measurements on-axis and applying the correct PEQ to smooth it in the areas it should be smoothed, which is what the perfectionist would do with not-good speakers, instead, the compromiser tries to make proxy backwards adjustments using ungated measurements and guessing on what ungated target curve might, hopefully, with luck, move things in the right direction, by proxy assumptions.

You clearly have no idea what I'm doing and are making untrue assumptions. You are embarrassing yourself.

 

[GalZohar]

That brings up a good point I should have mentioned--you always want to measure afterward to verify things line up correctly. With this flaw in the way Audyssey sets the levels and doesn't do the speaker and sub together to make sure they match at the crossover, you always want to be on the lookout for it so you can adjust if needed. Usually, if done as I suggest things do, but not always. Look for a response that follows the target curve through the crossover like so and adjust if necessary (an example of my left front speaker, crossover 60 Hz):

View attachment 235085


There could be many reasons for this. First, 10 dB is a lot. I believe the max boost Audyssey will do is 9 dB, so you'd need a fair amount of natural bass reinforcement to have a hope of getting that much. Secondly, the most likely reason it doesn't reach your intended bass boost is the LF Cutoff setting. If it's applying a second order high pass, etc, the speaker may never get all the way to the max boost you need. You may need to lower the frequency of it, change it to 4th order or override it completely to get that much boost.

My understanding is that Auydyssey sets the levels based upon an average from 500-2K Hz for the main channels, so the bass boost below that shouldn't affect the level. However, as you may have noticed, changing the EQ limit above that frequency can. It's common for the speaker to have pretty large dips/peaks all the way past 500 Hz and as you move that curtain you are adding or subtracting them from the average which can affect the overall level. But it's usually within +/- 1 dB or so, so that wouldn't explain 10 dB. Keep in mind the EQ limit is not a brick wall--it continues to EQ an octave above where you set it, gradually reducing the magnitude of the filter adjustments, so moving around that EQ limit anywhere between 250-2K Hz can change the level a bit. But yes, EQing full range so you have a smooth response through that range generally takes care of the issue.

Out of curiosity I changed one of my old calibrations to have 10 dB of boost, here is the correction it shows with a limit of 300 and full range:

View attachment 235093

View attachment 235094

As you can see, it never gets to 10 dB, but does get close. Changing the EQ limit does change the level, but it's a small amount.


I haven't followed Arc for years, but at the time they were having all sorts of problems around issues like this. But that was years ago, I'd assume they have it fixed by now. With DLBC it certainly shouldn't be an issue with Dirac as it does the speaker/sub together for the channel which is a real advantage. I don't think the base level version does, but I haven't kept current on that either.

If you'd like to send me the file I'd be happy to take a look, you've got me curious.

It's not my PC so I don't have those files, but I reproduced the issue even with the 20$ app with a random measurement file.

The issue was confirmed by Audyssey support, without suggesting a workaround/solution.

Basically the total filter boost of the range you are EQing will always be 0. If you boost the entire range (doesn't matter by how much, I used 10db just as an example, but 1db would have same issue), you get zero boost. If you boost most of the range, you actually get a much lower boost, and then an appropriate cut on the other part of the range you EQ, for total of 0.

It's very easy to see, if you have a bass boost shelf filter well above the EQ limit frequency - You will get 0 boost. Remove the EQ limit frequency, and suddenly the boost is applied. This is possible to reproduce in both the 20$ mobile app as well as the 200$ MultEQ-X.

My suggested workaround would be to not actually limit the EQ frequency, but instead EQ full range, and just set all measurement values above the EQ limit frequency to as if they measured perfectly (could potentially smooth out the transition over an octave). That way any target curve changes above the EQ limit would work as a "manual EQ filter" (similar to what happens now if you delete all measurements completely), and below the EQ limit frequency you can still apply boosts relative to the entire range and not just relative to the range of the EQ-enabled frequencies. With my suggestion, it would get the "total of 0" by slightly lowering the levels at all other frequencies, including those in the manual-EQ range (above the EQ limit frequency). Theoretically, one might be able to write an external script/app to modify the MultEQ-X measurement files to apply this solution (without an update from Audyssey), but it wouldn't be a fun job to write and maintain such a script/app.

 

[Jon AA]

Are you talking about the Sub channel or other channels? They work differently. The issue you're describing is correct for the sub channel and what we have been talking about--if you use a curve for the sub with it boosted a certain amount on the low end, you generally need to increase the trim level by that amount after calibration. It doesn't work that way with speakers unless you're boosting up to a crazy high (for a speaker) frequency.

 

[Newman]

My old Yamaha AVR does not have RC or adjustable EQ functions; it only has a hardwired and non-adjustable loudness compensation, just like most conventional amps do. So it is rather improbable that it has any protective HPFs in the bass region.

I am willing to guarantee that it does. Because they all do. It is not a ‘deluxe nanny feature’ only found in higher end or more modern RC algorithms.

I bet you anything that, if you measure your speaker bass FR before and after RC with your old Yamaha, you will find that it does not add, for instance, 20 dB of boost where the ‘before‘ measurement shows bass rolled off by 20 dB.

Cheers

 

[KMO]

Are you talking about the Sub channel or other channels? They work differently. The issue you're describing is correct for the sub channel and what we have been talking about--if you use a curve for the sub with it boosted a certain amount on the low end, you generally need to increase the trim level by that amount after calibration. It doesn't work that way with speakers unless you're boosting up to a crazy high (for a speaker) frequency.

I saw the investigation of this somewhere else, and I was the one who suggested this "average gain of limited EQ band is zero" thing might be happening on main channel, mirroring what the sub does. (Perhaps because the algorithm got reused...)

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

In my experience, Audyssey detects and does not attempt to boost null regions of the response. It will fill in the edges and make it narrower, which helps make it less audible.

www.audiosciencereview.com

On Galz's investigation, it really did seem to be the case.

So when you say "It doesn't work that way with speakers", are you just asserting this, or have you done some investigation of your own? Can you show some measurements showing an overall boost to the limited EQ range?

It may be there's some complicating factor, like there being some limit to the overall boost, which if exceeded snaps the overall boost back to zero, rather than clamping?