The problem with Audyssey Dynamic EQ

[-Matt-]

Here is what Audyssey does for me....

I have disproven your theory that Audyssey does not have a lot of resolution with measurements of my system over the years.

Indeed, these show nicely that the frequency response correction part of Audyssey XT32 is capable of working well, with adequate resolution at low frequencies.

Perhaps we could circle back now to the intended topic of the thread? ...Which is the way that Dynamic Eq has been implemented (especially in the surround channels).

 

[Chromatischism]

They may be using rectangular windowing with some tricky centring but still there's really a limit to what you can do with 1024 taps.

Again, in one of the Audioholics videos with Jeff Clark, he explains that the resolution is equivalent to 16,000 taps per channel, distributed per octave.

Because it varies by the spectrum you can't give a universal number, but he describes it as 1/12th octave at 40-50 Hz, or "note by note" accuracy.

What does this translate into as a frequency resolution at 20Hz?

He says it's about 1/5th octave at 20 Hz. That would be like 2 Hz resolution.

Here:

 

[Jon AA]

Dynamic EQ (0dB) boost starts from 1kHz and goes all the way down gradually increasing to 12.5dB boost over the Harman curve in sub bass area. This is way too loud by any measure and if it's improving your SQ, there must be a huge dip in your system's mid & bass response....

OK, multiple problems with this video.

First, DEQ is volume dependent. Absolute statements about what it does without mentioning the main volume setting are rather meaningless. Here are some measurements I took years ago (please ignore the absolute SPL levels--back then I was using an outboard soundcard and obviously forgot to calibrate the absolute mic level):

As you can see, at a low volume level DEQ adds a TON of bass. At a higher level, it adds a much more moderate amount. In fact, with the reference level set to -15 and the main volume set to -15, DEQ does absolutely nothing to your main channels. If you measure at full reference level, you'll find DEQ only will subtract from your bass level.

Here is another set of measurements I took at -20 MV:

Secondly, very, very few people after an Audyssey calibration end up with a large boost in the bass as you have unless they crank up their sub level afterward (which is very common to do). Most are left with very flat bass in which case the DEQ needs to add quite a bit just to get up to the level of the Harman curve you show. It's pretty widely known that if you crank up your sub level after calibrating, you may want to minimize the use of the DEQ as the two combined may be too much.

Personally, I have my target curve set with the bass boost wanted already in it and so when I do use the DEQ I typically keep it at -15 Ref setting since most of the time when I care about sound quality I average around a -15 main volume level--which means it's doing nothing most of the time to the main channels. Other than that, it can be a useful/quick/easy "tone control" by adjusting the Ref Level Offset.

Though I do agree and am annoyed by the change to the surround level as Matt mentioned (to get back to the subject of the thread) and would really like it if they changed that.

And no, the sub isn't causing the high frequency slope/rolloff in your measurements. That's quite impossible as the highest low-pass one can have on a D&M is 250 Hz. The sub is always low-passed.

As to your contention the base-level Audyssey amounts to an FIR with only two taps, that's quite unbelievable and unsupported, especially when Audyssey is saying something much different. It has been so long since I used the base level version I don't think I even had measuring equipment back then so I have no measurements to share, but I do remember seeing plenty of others that would show that clearly wasn't the case.

I was quite astounded just how much resolution XT32 was capable of when I started manually shaping it with control points using the App. These days with MultEQ-X, they let you see it. Here's an example:

And yes, all those little squiggles actually are doing something. I don't know enough about FIR filters to eyeball that and estimate the equivalent number of taps one would need to duplicate it with other software, so I'll leave those estimations to others.

 

[Chromatischism]

As to your contention the base-level Audyssey amounts to an FIR with only two taps, that's quite unbelievable and unsupported, especially when Audyssey is saying something much different. It has been so long since I used the base level version I don't think I even had measuring equipment back then so I have no measurements to share, but I do remember seeing plenty of others that would show that clearly wasn't the case.

I don't have access to the base level, but Audyssey stated long ago that it was "dozens", XT was "hundreds", and XT32 was "more than ten thousand" points. That was 10 years ago. It was more recently revealed that number for XT32 is around 16,000.

Thanks for helping to clear up the misinformation.

 

[OCA]

OK, multiple problems with this video.

First, DEQ is volume dependent. Absolute statements about what it does without mentioning the main volume setting are rather meaningless. Here are some measurements I took years ago (please ignore the absolute SPL levels--back then I was using an outboard soundcard and obviously forgot to calibrate the absolute mic level):

View attachment 228799

As you can see, at a low volume level DEQ adds a TON of bass. At a higher level, it adds a much more moderate amount. In fact, with the reference level set to -15 and the main volume set to -15, DEQ does absolutely nothing to your main channels. If you measure at full reference level, you'll find DEQ only will subtract from your bass level.

Here is another set of measurements I took at -20 MV:

View attachment 228800

Secondly, very, very few people after an Audyssey calibration end up with a large boost in the bass as you have unless they crank up their sub level afterward (which is very common to do). Most are left with very flat bass in which case the DEQ needs to add quite a bit just to get up to the level of the Harman curve you show. It's pretty widely known that if you crank up your sub level after calibrating, you may want to minimize the use of the DEQ as the two combined may be too much.

Personally, I have my target curve set with the bass boost wanted already in it and so when I do use the DEQ I typically keep it at -15 Ref setting since most of the time when I care about sound quality I average around a -15 main volume level--which means it's doing nothing most of the time to the main channels. Other than that, it can be a useful/quick/easy "tone control" by adjusting the Ref Level Offset.

Though I do agree and am annoyed by the change to the surround level as Matt mentioned (to get back to the subject of the thread) and would really like it if they changed that.

And no, the sub isn't causing the high frequency slope/rolloff in your measurements. That's quite impossible as the highest low-pass one can have on a D&M is 250 Hz. The sub is always low-passed.

As to your contention the base-level Audyssey amounts to an FIR with only two taps, that's quite unbelievable and unsupported, especially when Audyssey is saying something much different. It has been so long since I used the base level version I don't think I even had measuring equipment back then so I have no measurements to share, but I do remember seeing plenty of others that would show that clearly wasn't the case.

I was quite astounded just how much resolution XT32 was capable of when I started manually shaping it with control points using the App. These days with MultEQ-X, they let you see it. Here's an example:

View attachment 228804

And yes, all those little squiggles actually are doing something. I don't know enough about FIR filters to eyeball that and estimate the equivalent number of taps one would need to duplicate it with other software, so I'll leave those estimations to others.

I agree with you that the video measurement was not done at reference level and DEQ effect would be less if I had measured at a higher level. But I must say that the level was as high as I can listen to my system, above that I get neighbours knocking the walls.

"Dynamic" EQ should be volume dependent by definition however 20 dB (if you take flat response instead of Harman as ref) amplification on the sub bass is too much. I have compared it with Fletcher-Munson curves after @-Matt- 's warning and still found that it over compensates 5-6 dB at levels little below the ref levels. More importantly there are no SPL levels or no DEQ levels at which the room response follows Harman curve perfectly. Justifying it by some people cranking up sub volumes, etc. does not fit with Occam's Razor principle IMO.

I also disagree with your claim that sub is completely cut-off at 250Hz. Here's the evidence (psychoacoustic smoothing applied to all):

50dB continuous noise all the way to nearly 1000Hz, never falling below 43 dB and even rising again from 10kHz onwards! Even considering room noise threshold around 40 dB (and it was less than that in the room during this measurement) this is quite something. And you can clearly see the effect on Left speakers SPL when you vector average them which takes into consideration phase responses as well.

@amirm has also spotted the drop in HF during his test of Audyssey back in Apr 2020 https://www.audiosciencereview.com/forum/index.php?threads/audyssey-room-eq-review.12746/) and has then said: "The big dip at the end is probably due to my measurement mic running without calibration." But I measured these with a calibrated Umik-2!

This is not Audyssey's fault though. It's the crossover quality (or the lack of it) of the receiver. My Marantz SR6011 didn't have these residuals, the new SR6015 has. They've added IMAX though

 

[Chromatischism]

UMIK-2 will be worse at measuring high frequencies in-room due to its larger capsule.

With that said, if that is a Reference curve measurement, it looks to match theirs pretty well. I'll be the first to say I'm not a fan of it because the trendline is carried too far out before dropping. The slope should be more gentle and begin sooner.

Also, your sub measurement makes no sense, and looks like an anomaly of some kind. You may be measuring your room's noise floor.

DEQ was invented long before the Harman curve. However, with DEQ I normally end up in the ballpark anyway.

My system does not see a 20 dB increase in the bass. The typical amount is 8-10 dB from 20-40 Hz at -20 to -25 MV.

 

[abdo123]

I don't like the Harman target curve bass boost at all to be honest. What i usually use in my system is a small tilt from 100Hz downward (2db/oct).

For me a fixed boost in bass is really over the top in anything that is mixed to be peaking over 80dBSPL. It stops me from enjoying higher playback levels in movies.

 

[OCA]

Green line is Harman target curve, blue line is my actual left speaker response with MultEQ on, everything else off (I use a heavily edited curve in MultEQ app), red line is Audyssey reference calibration. yellow line are the changes DEQ would do.

It only corrects the high HF properly. Yes you can keep sub bass low so DEQ corrects it to the right level but the already high 150-800 range will always be boosted even more with DEQ thus it should be avoided.

These are accurate REW measurements, not my ideas. Audyssey boosts midrange for every speaker similarly. I have hundreds (if not thousands) of measurements in many different rooms over the years. Feel free to share your actual response curves and let's compare. I claim that you can never get anywhere close to my speaker response at any volume level with DEQ on.

PS measurements are FDW:15, no smoothing

 

[abdo123]

View attachment 228917
Green line is Harman target curve, blue line is my actual left speaker response with MultEQ on, everything else off (I use a heavily edited curve in MultEQ app), red line is Audyssey reference calibration. yellow line are the changes DEQ would do.

It only corrects the high HF properly. Yes you can keep sub bass low so DEQ corrects it to the right level but the already high 150-800 range will always be boosted even more with DEQ thus it should be avoided.

These are accurate REW measurements, not my ideas. Audyssey boosts midrange for every speaker similarly. I have hundreds (if not thousands) of measurements in many different rooms over the years. Feel free to share your actual response curves and let's compare. I claim that you can never get anywhere close to my speaker response at any volume level with DEQ on.

PS measurements are FDW:15, no smoothing

The boost (as broad as it is) is backed by research though.

 

[OCA]

The boost (as broad as it is) is backed by research though

In reality, you watch movies at one fixed level. Sometimes you cannot due to late night hours, baby sleeping, etc. but you then don't expect to hear massive explosions either. I calibrate my system for that level I always like to watch a movie and right on the harman curve which gives me flat perceived FR in the audible range.

Audyssey is not calibrating it to flat in the first place. it has odd dips at HF, odd boosts at midrange and dips again between 1500-7000 Hz (even more odd dips added if you use MRC as suggested). Btw the research (Fletcher-Munson equal loudness contour):

LF slope starts at 300Hz not 1kHz and HF should be boosted way up, DEQ can only bring it to flat at best (DEQ 0dB)

 

[Jon AA]

But I must say that the level was as high as I can listen to my system, above that I get neighbours knocking the walls.

Well, that's unfortunate. It's hard to really do any "critical listening" at such low levels. But if you must, then you should realize using DEQ at low volumes is going to add a ton of bass. If you don't like it for your use case, why not just switch it off?

I also disagree with your claim that sub is completely cut-off at 250Hz. Here's the evidence (psychoacoustic smoothing applied to all):

I meant for all practical purposes. Your measurement is not unlike mine:

Yes, some signal gets through. Like yours, mine is down around 45 dB. That's a lot. The increase over 10K for mine looks like it may be very worrysome! Until you look at the noise floor:

It's a non-issue. If you really think a signal -45 dB is having a large effect on you mains at high frequency, it's easy enough to take a measurement and without moving the mic, take another measurement with the sub amp shut off.

Audyssey boosts midrange for every speaker similarly. I have hundreds (if not thousands) of measurements in many different rooms over the years. Feel free to share your actual response curves and let's compare. I claim that you can never get anywhere close to my speaker response at any volume level with DEQ on.

Audyssey does what you tell it to do. When you have the app and like using the Harman curve, I'm not sure what all the complaining about the reference curve is accomplishing. It should be long forgotten. With a little effort, you can use Audyssey to hit any target curve you want (if your speakers and room are capable) if you have the app or MultEQ-X.

PS measurements are FDW:15, no smoothing

Why are you using windowing for these measurements to check what Audyssey is doing? It doesn't use windowing, it corrects the steady state response (as most room curves, including the Harman, are based upon steady state in room responses). If you want to correctly see if Audyssey is doing what it is told, you should use steady state measurements. A spatial average as well (Moving Mic Method covering a similar volume to the positions used for Audyssey calibration is the best way I've found to do it). It's unlikely a sweep at any single point in space is going to be a perfect match for the spatial average--which is why a spatial average is used.

 

[OCA]

Well, that's unfortunate. It's hard to really do any "critical listening" at such low levels. But if you must, then you should realize using DEQ at low volumes is going to add a ton of bass. If you don't like it for your use case, why not just switch it off?


I meant for all practical purposes. Your measurement is not unlike mine:

View attachment 228918

Yes, some signal gets through. Like yours, mine is down around 45 dB. That's a lot. The increase over 10K for mine looks like it may be very worrysome! Until you look at the noise floor:

View attachment 228919

It's a non-issue. If you really think a signal -45 dB is having a large effect on you mains at high frequency, it's easy enough to take a measurement and without moving the mic, take another measurement with the sub amp shut off.



Audyssey does what you tell it to do. When you have the app and like using the Harman curve, I'm not sure what all the complaining about the reference curve is accomplishing. It should be long forgotten. With a little effort, you can use Audyssey to hit any target curve you want (if your speakers and room are capable) if you have the app or MultEQ-X.

View attachment 228921


Why are you using windowing for these measurements to check what Audyssey is doing? It doesn't use windowing, it corrects the steady state response (as most room curves, including the Harman, are based upon steady state in room responses). If you want to correctly see if Audyssey is doing what it is told, you should use steady state measurements. A spatial average as well (Moving Mic Method covering a similar volume to the positions used for Audyssey calibration is the best way I've found to do it). It's unlikely a sweep at any single point in space is going to be a perfect match for the spatial average--which is why a spatial average is used.

You have the same sub SPL increase above 5K and it's a bit higher than the noise floor, also higher than NF up to 1000Hz, clearly the crossover is not really crossing over much. I hadn't checked the distortion graphs for NF, thanks for the idea. I am not sure of its effect, I agree that the volume is too low to change anything but phase cancellations can occur, REW vector average shows a real calculated dB drop. What I am sure of is that it's receiver related because I don't get it with the same mic, same position, same speaker with the stereo amp (a Naim).

I can comfortably say I am familiar with curve editing, the response I shared above is produced by over 50K custom points:

I am not complaining, you guys are complaining to keep the DEQ on. I am very happy with the current stage of my sound system That's why I am excited to share the methods.

Lastly, Audyssey is processing and getting rid of HF reflections, only not by frequency dependent windowing but by sending multiple chirps during measurement. The 3 extra chirps after the first one are just to eliminate reflections. The end results are supposed to be similar (though I cannot confirm this as it's being done in a black box in the receivers) but I can show you the actual Audyssey measurement (you need to FFT the MultEQ response data in Matlab to get that):

It's from 0 to 24kHz (hence 48 kHz sample rate), volume is 110dB, it has 16384 data points, 8 bit precision. In comparison, REW uses 131072 data points (8 times as much) and 32 bit floats for the same sample rate.

 

[OCA]

Guys, thanks to you, I have finally found out the main problem with Audyssey corrections. Chris Kyriakakis, the mind behind Audyssey happened to comment under my post in a fb group just a moment ago and I used the opportunity to ask him how they get rid of reflections by multiple chirps. He confirmed that multiple chirps are just to eliminate noise from measurements. The HF reflections are only smoothed out by 1/12 averaging. There's literally no FDW applied at any point.

As most of you will know, you correct ONLY for room reflections up to Schroeder frequency and you correct ONLY for direct sound at high frequencies. The way to achieve that is to get rid of reflections at HF by frequency dependent windowing. Otherwise you will falsely measure higher than actual SPL in HF boosted by all the room reflections. Look at the difference in HF for the same measurement below:

Red line is 1/12 smoothed, green line is variable windowed by frequency. It's not debatable that green line is the actual FR of that room. Audyssey is reading HF much higher than it really is, reading the reflections boosted SPL rather than the direct sound hence lowering HF to equalize. When based on the red response, obviously the corrected ref curve will end up too low in the HF. Any dynamic boost based on that will help things but will never be perfect.

Here I also added the original measurement with all reflections:

 

[abdo123]

As most of you will know, you correct ONLY for room reflections up to Schroeder frequency and you correct ONLY for direct sound at high frequencies. The way to achieve that is to get rid of reflections at HF by frequency dependent windowing. Otherwise you will falsely measure higher than actual SPL in HF boosted by all the room reflections. Look at the difference in HF for the same measurement below:

View attachment 228935
Red line is 1/12 smoothed, green line is variable windowed by frequency. It's not debatable that green line is the actual FR of that room. Audyssey is reading HF much higher than it really is, reading the reflections boosted SPL rather than the direct sound hence lowering HF to equalize. When based on the red response, obviously the corrected ref curve will end up too low in the HF. Any dynamic boost based on that will help things but will never be perfect.

Here I also added the original measurement with all reflections:

View attachment 228936

How do you do frequency dependent windowing in REW?

 

[OCA]

How do you do frequency dependent windowing in REW?

IR windows

 

[abdo123]

IR windows

How does it work? How is it frequency dependent?

 

[Jon AA]

Chris Kyriakakis, the mind behind Audyssey happened to comment under my post in a fb group just a moment ago and I used the opportunity to ask him how they get rid of reflections by multiple chirps. He confirmed that multiple chirps are just to eliminate noise from measurements. The HF reflections are only smoothed out by 1/12 averaging. There's literally no FDW applied at any point.

OK, now we're getting somewhere. It's nice that he responded so fast.

That's not the whole story though. If you're doing high frequency corrections, you don't want to try and fix sharp dips and peaks that are highly mic position dependent. The spatial average keeps you from doing that (has a smoothing effect). Here's a single MQX measurement compared with the aggregate (~25 mic positions or so) with all smoothing turned off:

You can see the dramatic smoothing effect a good spatial average gets you (and a big advantage MQX has by allowing so many measurements). With the lowest level of smoothing applied:

You can see they do apply smoothing before calculating the correction filter--which is a good thing, you certainly don't want to try and correct all that hash:

It looks like their filter is pretty much based on their middle smoothing setting, 1/12 looks about right:

And as we've now confirmed, it is based on the steady state response. Here's a comparison of the MQX aggregate to a MMM using a Umik1 for that speaker at the listening position:

The deviations at low frequencies are because I modified the speaker which reduced the sensitivity of the woofer at low frequencies. The deviations at high frequencies track well with the differences in calibration between the two mics (which I have measured and can compensate for in the target curve, but didn't for this particular file). Again I'll mention the number of measurements MultEQ-X allows you to take really helps you get a much better, repeatable spatial average.

OK, now that that's out of the way....

Otherwise you will falsely measure higher than actual SPL in HF boosted by all the room reflections. Look at the difference in HF for the same measurement below:
....It's not debatable that green line is the actual FR of that room. Audyssey is reading HF much higher than it really is, reading the reflections boosted SPL rather than the direct sound hence lowering HF to equalize.

It's not that one measurement is "correct" and the other is not, or that one is "true" and the other is not. You're measuring different things. Both are correct. And the Harman curve is based upon the red line, not the green one. If you correct a windowed response to that curve, you aren't really using that curve.

If you read Toole's book, all the papers, all the research from Harman, every time they refer to an "in room response" they are referring to the steady state response--the red line (unless specified otherwise). The Predicted In Room Response from the spin measurements is predicting the steady state response in a typical room. Good anechoic data can not really be obtained from your listening position in a room. I understand there are some newer/high end room correction programs that do some windowing, but I don't know enough about them to comment on how well they do at that. Historically, pretty much all room correction systems have used steady state measurements, and that's what all the various room curves people pass around are based upon.

As most of you will know, you correct ONLY for room reflections up to Schroeder frequency and you correct ONLY for direct sound at high frequencies.

Generally, that's the way most like to do it. But it's an oversimplification to say you can't do that with RC that uses steady state measurements. It takes some effort, knowledge and anechoic data from the speaker but it can be used for that purpose (it's just a lot more work that buying a JBL SDP-75 which allows you to load "Anechoic EQ"--which from those I've seen amounts to EQing the listening window flat) before you even start your in-room measurements.

Sorry for getting so off-topic, Matt. But I do think it's a useful discussion.

 

[OCA]

How does it work? How is it frequency dependent?

en:wiki:funktionen:td-functions:frequency_dependent_window [Acourate-Wiki]

www.audiovero.de

 

[Chromatischism]

View attachment 228917
Green line is Harman target curve, blue line is my actual left speaker response with MultEQ on, everything else off (I use a heavily edited curve in MultEQ app), red line is Audyssey reference calibration. yellow line are the changes DEQ would do.

It only corrects the high HF properly. Yes you can keep sub bass low so DEQ corrects it to the right level but the already high 150-800 range will always be boosted even more with DEQ thus it should be avoided.

These are accurate REW measurements, not my ideas. Audyssey boosts midrange for every speaker similarly. I have hundreds (if not thousands) of measurements in many different rooms over the years. Feel free to share your actual response curves and let's compare. I claim that you can never get anywhere close to my speaker response at any volume level with DEQ on.

PS measurements are FDW:15, no smoothing

Lol what is this. Your line is not even close to reality. All of the measurements I've posted had Dynamic EQ on with zero Reference Level Offset.

Here is another. Not a sine sweep, but an averaged moving mic at -20 MV:

Where is the boost that starts skyrocketing below 1 kHz?

You have the same sub SPL increase above 5K

It's your noise floor, not your speakers.

 

[OCA]

OK, now we're getting somewhere. It's nice that he responded so fast.

That's not the whole story though. If you're doing high frequency corrections, you don't want to try and fix sharp dips and peaks that are highly mic position dependent. The spatial average keeps you from doing that (has a smoothing effect). Here's a single MQX measurement compared with the aggregate (~25 mic positions or so) with all smoothing turned off:

View attachment 228967

View attachment 228968

You can see the dramatic smoothing effect a good spatial average gets you (and a big advantage MQX has by allowing so many measurements). With the lowest level of smoothing applied:

View attachment 228969

View attachment 228970

You can see they do apply smoothing before calculating the correction filter--which is a good thing, you certainly don't want to try and correct all that hash:

View attachment 228971

It looks like their filter is pretty much based on their middle smoothing setting, 1/12 looks about right:

View attachment 228972

And as we've now confirmed, it is based on the steady state response. Here's a comparison of the MQX aggregate to a MMM using a Umik1 for that speaker at the listening position:

View attachment 228974

The deviations at low frequencies are because I modified the speaker which reduced the sensitivity of the woofer at low frequencies. The deviations at high frequencies track well with the differences in calibration between the two mics (which I have measured and can compensate for in the target curve, but didn't for this particular file). Again I'll mention the number of measurements MultEQ-X allows you to take really helps you get a much better, repeatable spatial average.

OK, now that that's out of the way....


It's not that one measurement is "correct" and the other is not, or that one is "true" and the other is not. You're measuring different things. Both are correct. And the Harman curve is based upon the red line, not the green one. If you correct a windowed response to that curve, you aren't really using that curve.

If you read Toole's book, all the papers, all the research from Harman, every time they refer to an "in room response" they are referring to the steady state response--the red line (unless specified otherwise). The Predicted In Room Response from the spin measurements is predicting the steady state response in a typical room. Good anechoic data can not really be obtained from your listening position in a room. I understand there are some newer/high end room correction programs that do some windowing, but I don't know enough about them to comment on how well they do at that. Historically, pretty much all room correction systems have used steady state measurements, and that's what all the various room curves people pass around are based upon.



Generally, that's the way most like to do it. But it's an oversimplification to say you can't do that with RC that uses steady state measurements. It takes some effort, knowledge and anechoic data from the speaker but it can be used for that purpose (it's just a lot more work that buying a JBL SDP-75 which allows you to load "Anechoic EQ"--which from those I've seen amounts to EQing the listening window flat) before you even start your in-room measurements.

Sorry for getting so off-topic, Matt. But I do think it's a useful discussion.

Thank you for all the information. Especially because I don't have MQX, this has been quite useful. Spatial averaging seems to change the HF response but all locations will still have boosts albeit being caused by different reflections. The average FR will end up a bit randomly boosted. Spatial averaging needs to implement impulse correlation matrices to work properly. As far as I know Dirac also does not do that.

There's ongoing discussion on best windowing types or using different window widths for left & right depending on the actual room reverberation times (RT60), etc. for ideal FDW but noone questions its necessity.

I strongly suggest everyone to watch this quite long tutorial by Mitch Barnett. The guy has written a state of the art convolution engine capable of instantly switching between different impulses and is known to professionally provide successful room corrections to thousands of audiophiles: