Help finding amazing room correction post

[Snarfie]

Someone more clued up might set me right, but as I now understand it there’s a solid scientific reason not correct above 500Hz or so.

From what I’ve read in the posts quoting Toole, linked above, the sound heard at higher frequencies (above 500Hz or so) is a mixture of direct and reflected sound, but your brain filters out reflections as just reflections. However, a measurement mic can’t do that.

If you have speakers that measure flat under anechoic conditions, then measure those speakers in your room, your measurement mic will combine the direct and reflected sounds to show a response that is anything but flat.

It’s fine to correct lower frequencies as you’re removing the gain added by the room, which your brain can’t discern as anything else.

But if you correct the deviations from flat at higher frequencies, you’re adding filters to a response that your brain has already filtered, so you’re actually adding deviations from flat to the signal from your nice flat speakers.

I"m a beta tester at Mathaudio discussed with Sergey one of the technical contact persons my full range correction. He stated their nothing wrong regarding a full correction better he encourage it as a starting point. From there on move the slider up till maximum transparency specific for You is reached. So i did an because of the Scientific statement don't correctl not above 4 or 500 hz I let many people hear specific in my room the correction bypass sound an other target curves basicly all agree the full range correction is by far the best nothing to do with your mind pure observation.
I guess that Mathaudio has a specific way of correcting. Because my Vandersteen speakers are by build phase coherent speakers I did ask how Mathaudio handels phase Behaviour Mathaudio did not want to eleborate in depth about how they do it (Competively quite understandable) other than if i have phase coherent speakers Mathaudio has to correct less.

See graph from Mathaudio as a starting point. Way down (in contrast to only correct till 4 or 500hz) Full correction an work you way up if necccecery. https://mathaudio.com/room-eq.htm

 

[Jas0_0]

You know, that's like saying room treatment and acoustical diffusers, rugs. etc. don't matter.
Do not take everything from one person as the only truth. Like with most things it is not black or white.

If that statement was true, you could just place your speakers in a bare naked room with concrete walls and hard-wood floor and it would be no different sounding to your ears than if you treated the room.. Because your brain filters out the reflections!?

Sorry, I'm not believing it. And I correct full range by the way.

Perhaps my explanation lacked clarity. Rugs, room treatment etc really do matter.

I wasn’t suggesting that your brain can remove the reflected sound, more that it understands that reflections are reflections (in a way that a mic and algorithm doesn’t) and that’s why using EQ to try to remove their impact doesn’t work.

Room treatment helps filter out reflections. Whereas EQ filters the direct sound and resulting reflections. So you don’t get better sound, you just get a signal that is still made up of wanted direct and unwanted reflections that’s now louder and quieter at certain frequencies.

 

[Snarfie]

Problems without (full) correction that i face is as follows. Analyzing my frequency response (yes i did with several measuring mic's to be sure) is that i have Up to 14db difference this is really huge. The uncorrect sound makes mid en high frequencies dominating bass is basicly gone. A other problem because of the mid en high frequencies there is an hiss between voices instrument causing lack the sound of depth imaging. See it as the old fashion tone control (bass, trebel) which is a maximum of let 3 a 4 db increase or decrease in a certain frequency range. In my frequency curve it some time 5x that much causing instant fatigue. So correction only till 500hz is really no option. Could you imagen what happend when i corrected these anomalies with Mathaudio basicly from one on the other moment i got a total balanced sound did rediscoverd my whole audio collection. My advise is to try at least for 1/2 hour full correction (to get ride of acoustics addictions.) an than use the bypass button see what happens an go from there on.

 

[Jas0_0]

Problems without (full) correction that i face is as follows. Analyzing my frequency response (yes i did with several measuring mic's to be sure) is that i have Up to 14db difference this is really huge. The uncorrect sound makes mid en high frequencies dominating bass is basicly gone. A other problem because of the mid en high frequencies there is an hiss between voices instrument causing lack the sound of depth imaging. See it as the old fashion tone control (bass, trebel) which is a maximum of let 3 a 4 db increase or decrease in a certain frequency range. In my frequency curve it some time 5x that much causing instant fatigue. So correction only till 500hz is really no option. Could you imagen what happend when i corrected these anomalies with Mathaudio basicly from one on the other moment i got a total balanced sound did rediscoverd my whole audio collection. My advise is to try at least for 1/2 hour full correction (to get ride of acoustics addictions.) a than use the bypass button see what happens an go from there on. This is what i do to people that never heard my system an have to qualify.

According to the science wot I just learned (and have now become a little evangelical about), there is nothing wrong with broad band tone control-type filters to correct overall sound balance.

The problem is when a room correction algorithm makes narrow filters to try to correct room-generated narrow deviations from flat at higher frequencies.

 

[Snarfie]

According to the science wot I just learned (and have now become a little evangelical about), there is nothing wrong with broad band tone control-type filters to correct overall sound balance.

The problem is when a room correction algorithm makes narrow filters to try to correct room-generated narrow deviations from flat at higher frequencies.

I'm with you regarding science thats why i'm on ASR brodening my knowledge. Point is that we really don't know in detail how the different room correction software correct the problems like phase time domain for instance. Why would Mathaudio advice to start with a Full correction en go from there on? Most important trust your own ears despite science. Would say give it a try full correction at least for 15 to 30 minutes an see what happens nothing to loose.

 

[Deleted member 48726]

I'm with you regarding science thats why i'm on ASR brodening my knowledge. Point is that we really don't know in detail how the different room correction software correct the problems like phase time domain for instance. Why would Mathaudio advice to start with a Full correction en go from there on? Most important trust your own ears despite science.

And Dirac as well.

 

[Snarfie]

And Dirac as well.

Interesting you use full correction in Dirac ?

 

[Deleted member 48726]

Interesting you use full correction in Dirac ?

Absolutely. Sounds the best. House curve applied full range.

 

[Snarfie]

Absolutely. Sounds the best. House curve applied full range.

Ha ha so it is not only Mathaudio. Those RC software compagnies are very reluctant to explain how the build there algo's quite understandable. Friend of mine use Dirac to will ask him what he uses.

 

[dominikz]

Problems without (full) correction that i face is as follows. Analyzing my frequency response (yes i did with several measuring mic's to be sure) is that i have Up to 14db difference this is really huge. The uncorrect sound makes mid en high frequencies dominating bass is basicly gone. A other problem because of the mid en high frequencies there is an hiss between voices instrument causing lack the sound of depth imaging. See it as the old fashion tone control (bass, trebel) which is a maximum of let 3 a 4 db increase or decrease in a certain frequency range. In my frequency curve it some time 5x that much causing instant fatigue. So correction only till 500hz is really no option. Could you imagen what happend when i corrected these anomalies with Mathaudio basicly from one on the other moment i got a total balanced sound did rediscoverd my whole audio collection. My advise is to try at least for 1/2 hour full correction (to get ride of acoustics addictions.) an than use the bypass button see what happens an go from there on. This is what i do to people that never heard my system an have to qualify.

To be honest, @Snarfie your uncorrected response looks like nothing I've seen before - the difference in level between <300Hz and above is really strange and makes me suspect something is wrong with the loudspeakers (or somewhere else in the playback system). This is why I'm not surprised some kind of full-range correction is mandatory for your system. But note that yours is a very peculiar case and therefore not necessarily applicable to the many playback systems that measure closer to 'normal'.

E.g. in my two systems (one with Revel M16 + SVS SB-1000, and the other with Neumann KH120A with Adam Sub7) there is IMHO definitely no need for full-range correction. When I tested full-range correction in my systems it was at best similar to correcting below the transition frequency only, and at worst it was significantly worse than just correcting below the transition frequency. This is because these loudspeakers measure well anechoically (they have 'good' spinoramas).

Also, saying that "science says" not to apply EQ full-range is IMO a misrepresentation. What is really meant is that if we use well-measuring loudspeakers we shouldn't need to apply EQ above the transition frequency. The reason is that humans appear to be able to distinguish between loudspeaker response resonances and room reflection effects above the room transition frequency, while room correction algorithms based on in-room steady-state response cannot - meaning that there is a danger of making things worse if those are applied above the transition frequency.

Note that this doesn't mean that you get the same sound with the same loudspeakers in a treated vs untreated room - it just means that we are able to hear 'bad' rooms and 'bad' loudspeakers above the transition frequency separately. If you have a 'bad sounding' room above the transition frequency you need to treat the room, and not expect room EQ to fix it.

Similarly, in case we have a poor measuring loudspeaker applying EQ above the transition frequency can absolutely be beneficial (and especially so if the loudspeakers have even directivity) - but in that case the best course of action is to EQ that region based on complete anechoic spinorama-style measurements. I.e. this is 'loudspeaker correction' and not 'room correction' anymore.
In case complete anechoic loudspeaker measurements are not available it is IMHO absolutely valid to try full-range correction with any kind of room EQ and hope for the best. But this is really the last course of action, with results that are difficult to predict (i.e. it will be better in some cases and worse in others).

 

[dominikz]

@Snarfie Perhaps I should also add that I believe that if you're happy with how your system is setup that is absolutely great and you should keep enjoying it as is. There is no need to change anything if you're happy - after all, the whole point of the hobby should be to enjoy the sound and the music.

But we should IMO be careful with advice we give to others - I personally try to steer people towards what established researchers in the field advise, as that is IMHO the most universally applicable advice.

 

[Snarfie]

To be honest, @Snarfie your uncorrected response looks like nothing I've seen before - the difference in level between <300Hz and above is really strange and makes me suspect something is wrong with the loudspeakers (or somewhere else in the playback system). This is why I'm not surprised some kind of full-range correction is mandatory for your system. But note that yours is a very peculiar case and therefore not necessarily applicable to the many playback systems that measure closer to 'normal'.

E.g. in my two systems (one with Revel M16 + SVS SB-1000, and the other with Neumann KH120A with Adam Sub7) there is IMHO definitely no need for full-range correction. When I tested full-range correction in my systems it was at best similar to correcting below the transition frequency only, and at worst it was significantly worse than just correcting below the transition frequency. This is because these loudspeakers measure well anechoically (they have 'good' spinoramas).

Also, saying that "science says" not to apply EQ full-range is IMO a misrepresentation. What is really meant is that if we use well-measuring loudspeakers we shouldn't need to apply EQ above the transition frequency. The reason is that humans appear to be able to distinguish between loudspeaker response resonances and room reflection effects above the room transition frequency, while room correction algorithms based on in-room steady-state response cannot - meaning that there is a danger of making things worse if those are applied above the transition frequency.

Note that this doesn't mean that you get the same sound with the same loudspeakers in a treated vs untreated room - it just means that we are able to hear 'bad' rooms and 'bad' loudspeakers above the transition frequency separately. If you have a 'bad sounding' room above the transition frequency you need to treat the room, and not expect room EQ to fix it.

Similarly, in case we have a poor measuring loudspeaker applying EQ above the transition frequency can absolutely be beneficial (and especially so if the loudspeakers have even directivity) - but in that case the best course of action is to EQ that region based on complete anechoic spinorama-style measurements. I.e. this is 'loudspeaker correction' and not 'room correction' anymore.
In case complete anechoic loudspeaker measurements are not available it is IMHO absolutely valid to try full-range correction with any kind of room EQ and hope for the best. But this is really the last course of action, with results that are difficult to predict (i.e. it will be better in some cases and worse in others).

As i stated before i did measure at least 3 or 4 different colum speakers same size all have more or less the same response. Elipson 1303, B&W, Optima 3 JK acoustics, Vandersteen model 1 . Also checked my measuring mic did measurment with 3 difference mic's to be sure. You more than welcome to listen if you are in the neighbourhood (Germany is not that far from Holland ). Looking to both speakers almost identical low mid high frequency response So something wrong with either speakers or the measuring is not likely. As suggested by Mathaudio to start with a full correction (see there website https://mathaudio.com/room-eq.htm ) what are you thoughts regarding that advise.

 

[Snarfie]

@Snarfie Perhaps I should also add that I believe that if you're happy with how your system is setup that is absolutely great and you should keep enjoying it as is. There is no need to change anything if you're happy - after all, the whole point of the hobby should be to enjoy the sound and the music.

But we should IMO be careful with advice we give to others - I personally try to steer people towards what established researchers in the field advise, as that is IMHO the most universally applicable advice.

I'm with your approche i only don't have an idea what the different algo's from different software brands are correcting exactly in detail and how they display the result. For instance about phase correction got the impression from you that Mathaudio does not do much regarding correcting that but Mathaudio do on there website. When i ask them how they do it i don't get answers understandable i guess.

 

[dominikz]

Point is that we really don't know in detail how the different room correction software correct the problems like phase time domain for instance.

What do you think is unknown? It is easy to test filters generated by various room correction tools - e.g. I did that in the thread linked in my signature.
Most room EQ SW correct phase together with magnitude in very similar ways because most of them use minimum phase filters.
Some allow use of linear phase filters (e.g. IK Multimedia ARC System 3 or Sonarworks Reference 4 Studio) which then doesn't correct phase (only magnitude).
Dirac Live additionally corrects for crossover phase wrap by using all-pass filtering. Acourate and JuiceHiFi Audiolense XO are able to do this as well. Unfortunately audibility of this is often disputed - see this article.

I'm without approche i only don't have an idea what the different algo's from different software brands are correcting exactly in detail. For instance about phase correction got the impression from you that Mathaudio does not do much regarding correcting that but Mathaudio do on there website. When i ask them how they do it i don't get answers understandable i guess.

MathAudio Room EQ (when I tested it) used standard minimum phase filters - IMO nothing out of the ordinary, really:

(Link1 and Link2 to my MathAudio tests)
So it 'corrects' phase as much as any other minimum-phase EQ with the same shape does (even simple IIR PEQ 'corrects' phase in this sense).

As i stated before i did measure at least 3 or 4 different colum speakers same size all have the same response. Elipson 1303, B&W, Optima 3 JK acoustics, Vandersteen model 1 all more or less the same response. Also checked my measuring mic's did measurment with 3 difference mic's to be sure.

Looking to both speaker almost identical low mid high frequency response So something wrong with either speakers or the measuring is not likely.

It is great that you did some additional tests. I guess there could also be issues upstream in the playback chain, or possibly in part even due to loudspeaker/listening position placement and room acoustics, etc; it is very difficult to say exactly why without looking into it very closely and doing many measurements.
However, your in-room response is IMO still very atypical - you can see it right away if you compare to basically any in-room response on this forum. Here's mine with Revel M16 and MathAudio EQ (vertical mic orientation with 90° mic calibration):

You more than welcome to listen (Germany is not that far from Holland ).

Thanks a lot for the offer, appreciated, but I'm not in Holland nor very close to Germany.
Again, as I said before, the full-range correction might work fine in your case; unfortunately that doesn't mean it is universally good advice for others.

As suggested by Mathaudio to start with a full correction (see there website https://mathaudio.com/room-eq.htm ) what are you thoughts regarding that advise.

MathAudio is obviously designed for full-range correction and the instructions are aligned with that. The author surely had his reasons to do it like that (I imagine it might have to do it with ease of use for consumers and the expectation that many PC systems have terrible loudspeakers; but who knows?), however that alone doesn't mean it is the ideal approach in every scenario.

 

[Snarfie]

What do you think is unknown? It is easy to test filters generated by various room correction tools - e.g. I did that in the thread linked in my signature.
Most room EQ SW correct phase together with magnitude in very similar ways because most of them use minimum phase filters.
Some allow use of linear phase filters (e.g. IK Multimedia ARC System 3 or Sonarworks Reference 4 Studio) which then doesn't correct phase (only magnitude).
Dirac Live additionally corrects for crossover phase wrap by using all-pass filtering. Acourate and JuiceHiFi Audiolense XO are able to do this as well. Unfortunately audibility of this is often disputed - see this article.

MathAudio Room EQ (when I tested it) used standard minimum phase filters - IMO nothing out of the ordinary, really:

(Link1 and Link2 to my MathAudio tests)
So it 'corrects' phase as much as any other minimum-phase EQ with the same shape does (even simple IIR PEQ 'corrects' phase in this sense).


It is great that you did some additional tests. I guess there could also be issues upstream in the playback chain, or possibly in part even due to loudspeaker/listening position placement and room acoustics, etc; it is very difficult to say exactly why without looking into it very closely and doing many measurements.
However, your in-room response is IMO still very atypical - you can see it right away if you compare to basically any in-room response on this forum. Here's mine with Revel M16 and MathAudio EQ (vertical mic orientation with 90° mic calibration):

Thanks a lot for the offer, appreciated, but I'm not in Holland nor very close to Germany.
Again, as I said before, the full-range correction might work fine in your case; unfortunately that doesn't mean it is universally good advice for others.

MathAudio is obviously designed for full-range correction and the instructions are aligned with that. The author surely had his reasons to do it like that (I imagine it might have to do it with ease of use for consumers and the expectation that many PC systems have terrible loudspeakers; but who knows?), however that alone doesn't mean it is the ideal approach in every scenario.

As you stated if you are happy thats probably the most important thing aiming at.

Regarding my new found sound let me put it like this. Because i correct full i got ride of the hiss (sometimes close to 14db) between voices and instruments that added considerably in a absolute better balanced sound.
An yes i got sometimes the impressions that voices are less natural sounding or instruments lacking a bit of clarity. Basicly room correction is by far not perfect it is more a trade off to get IMO the
most balanced sound not breaking the bank by buying new expensive gear or more expensive room treathment. Room correction software was an is a eye opener. Keep up the good work.

 

[Deleted member 48726]

Dirac suggests FR correction as standard. You can however drag the curtains from each end of the spectrum as you please.

 

[Jas0_0]

I think I’d be more worried about what’s causing 14db of hiss than room correction.

As @dominikz suggested, is something not quite right with your system?

Is the amp playing up, or some setting in the audio chain artificially boosting treble?

 

[Snarfie]

I think I’d be more worried about what’s causing 14db of hiss than room correction.

As @dominikz suggested, is something not quite right with your system?

Is the amp playing up, or some setting in the audio chain artificially boosting treble?

I'm not going repeat what i measured regarding the system to count that out. The amp a NAD C370 is using the bypass knob regarding tone controll. The amp is fine finding the new found sound an not only by me. The attic has a odd shape like a saddel roof that basicly result in a considerable less squere meters. So probably also strange room acoustics.

 

[goat76]

I'm without approche i only don't have an idea what the different algo's from different software brands are correcting exactly in detail and how they display the result. For instance about phase correction got the impression from you that Mathaudio does not do much regarding correcting that but Mathaudio do on there website. When i ask them how they do it i don't get answers understandable i guess.

To be frank, I think it's a business decision that firms like Mathaudio, Dirac, and Audyssey that do automatic room corrections need to sell you the idea that full corrections are in general a good thing to do. From their standpoint, the whole idea is that it's a fairly straightforward process, just put up the microphone at the main listening position and then the secondary ones, and the program will make corrections to the full frequency response and not just the bass region which could be seen as a lesser value considering it's such a small part of the full frequency range.

What they should do instead, which of course would destroy the straightforward process (the business strategy of an easy-to-use program), is they should give the users instructions on how to make gated measurements of their loudspeakers, which then the program should base the corrections for, from around 500Hz to 20kHz. The problem with such an approach is that it would make the whole process way more complicated (and would therefore likely be seen as too high of a step for the average consumer), as the users must be able to put both their speaker and the microphone as far from the room boundaries to get a large reflection-free measuring window as possible, which in turn would open up for many potential user errors.



As was mentioned by @dominikz, your situation seems to be an extreme case so the overall improvements of the corrections of the full frequency response are likely way better than no corrections at all. But it's still likely that corrections based on gated measurements above 500Hz, and room-based corrections under 500Hz would give you an even better result.

And even in less extreme situations than yours, I think people who like the sound of full corrections (based completely on none-gated measurements), mostly hear the improvements done to the bass region (which in most cases will give the largest improvement) and based on that overall impressions take it as granted that the full correction was the right thing to do, even if it's possible that some degrading may have come "with the package" of full corrections based on none-gated measurements.

My suggestion is that you figure out how you do gated measurements, make the corrections above 500Hz based on that, and leave the corrections under 500Hz to Mathaudio. I think that will improve your sound even further.
That is of course still "full range corrections", just that the corrections are based on the room response under 500Hz, and the rest are based on the direct response of your loudspeakers.


P.S. This post is not meant to upset anyone who has full belief in the firms that make automatic full-range correction programs, it's just that people like Toole who have done deep research in the field, have concluded that what the human sense of hearing "picks up" from the listening position is not the same as what a microphone picks up from the same spot, which is pretty clear if you just listen to the recorded signal and compare it to what you hear from the same spot.

 

[DDF]

FWIW
Room EQ FAQ