• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

EQing in stereo vs Eqing in mono

sfdoddsy

Active Member
Joined
Jun 6, 2019
Messages
293
Likes
438
Like most (I believe) I have always measured my speakers one channel at a time and then EQed that channel as needed.

This holds true whether I do it manually via REW and a MiniDSP, or automatically via Anthem ARC (or Dirac/Audyssey/whatever).

However, this article makes a considerable amount of sense:

https://www.tonmeister.ca/wordpress...e-room-compensation-some-more-details-part-2/

I'm wondering, however, how one puts the theory to the test with the above tools at my disposal.

I assume one would abandon automatic EQ systems. Unless one exists that does measure all your sources together.

Manually, you can measure L+R (with bass management) in REW, but the MiniDSP can only EQ per channel. So would you have to measure, tweak a channel, measure, tweak another channel, over and over until you get the combined response you're after.

And how high up do you go?
 
  • Like
Reactions: jlo

RayDunzl

Grand Contributor
Central Scrutinizer
Joined
Mar 9, 2016
Messages
13,202
Likes
16,982
Location
Riverview FL
Manually, you can measure L+R (with bass management) in REW, but the MiniDSP can only EQ per channel. So would you have to measure, tweak a channel, measure, tweak another channel, over and over until you get the combined response you're after.

Wouldn't you just measure L+R, create a filter for that response, and apply the filter to both channels on the miniDSP?

---

Here is the filter response I have (had) here for left and right filters (red), and the "vector average" of them (black), which would represent the "mono" filter, just to visualize the difference:

(the choice was for "sharp" filtering, "fix all the wiggles", as opposed to a "lazy" filter, which would do more averaging and create a "smoother" filter. The filter uses both FIR - for the wiggles and phase - and IIR - for the broad strokes)

1604100375602.png


And the phase adjustment:

1604100544765.png


Without a filter applied, both of the above traces would be a flat line - measurement taken at the preamp output.


I assume one would abandon automatic EQ systems. Unless one exists that does measure all your sources together.

My automated measure (AcourateDRC) does measure left and right separately, but I have the ability to make the signal mono so left or right would go to both speakers.

I've thought about doing that, but haven't.


And how high up do you go?

All the way.
 
Last edited:

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
I often measure both L+R channels at the same time in REW. However, you need to exactly align the microphone in between speakers, that is if you're doing it the hard way i.e. manually adjusting distances and not relying on automatically calculated delays.

With regards to shared equalization between speakers...

This really depends on how symmetrical the response is as measured in your room. Above the transition zone (speaker dominated region) you can use the same EQ for both speakers -- if you are sure they are well matched.
 

Vasr

Major Contributor
Joined
Jun 27, 2020
Messages
1,409
Likes
1,922
That line of thinking is flawed because in any stereo content, most of the material is skewed towards one or the other than being in the center. So, L+R correction will under-correct or over-correct for all of that content. L+R correction works best when you have only mono going through both speakers (which is what you are doing when you feed a signal for measurement).

There seems to be several logical flaws in that article:

1. I don't where the assumption that bass frequencies are always in the "center" equally in both speakers comes from. First of all if you have a sub, that is obviously false. Second, the bass frequencies are where the instrument producing the sound is in a good mix designed to present a sound stage. Not because it is directional, but typically it is not just the fundamental that one hears but all the natural harmonics as well harmonics from several notes played together on that same instrument. Why would anyone separate out the lower frequencies and re-mix it to both speakers in the recording away from the "effective location" of the rest of the directional harmonic content if that is likely to create phase difference between them. There might be some other technical reasons that I am not aware of but that looks like a false premise.

2. The best listening position is typically at the same distance from both speakers and if the both the speakers are aligned correctly in phase, the sound from both will arrive in phase at the listening position. What happens to the phase at each of the speaker locations relative to each other as used in the examples is irrelevant to the listener. So I don't get that logic.

3. I am not sure I buy the "music recording itself may be misaligned in phase". If it is due to individual recording idiosyncracies, which I find hard to believe, then every recording will different types of phase alignments so measuring once with phase aligned signals and correcting it makes no sense.

Seems like a very half-boiled, ill-thought-out argument to me. But I could be wrong. :)
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
Manually, you can measure L+R (with bass management) in REW, but the MiniDSP can only EQ per channel. So would you have to measure, tweak a channel, measure, tweak another channel, over and over until you get the combined response you're after.

There's a "link' checkbox on miniDSP which allows you to share EQ settings for a pair of channels. There's also a separate input stage and output stage area so if you add a PEQ at 60Hz in your input stage, your sub out channel will receive the modified signal... actually, you can pretty much put all your sub EQs in the input stage and just leave the sub channel output settings blank -- except for the xo -- same goes for the speaker out channels.
 

Sir Sanders Zingmore

Addicted to Fun and Learning
Forum Donor
Joined
May 20, 2018
Messages
947
Likes
1,898
Location
Melbourne, Australia
Silly question, perhaps. If you are generating PEQ settings by measuring each speaker individually, does that mean you have a separate set of PEQs for each channel? Or do you combine the measurements somehow before generating the EQ filters?
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
Skimmed the article... eh, I don't really think that theoretically deep about it. So I'll leave that to others.

All I can say is that I use both -- summed and mono.

The ff. is just an overview sample of how I went about with my manual bass managed EQ corrections in one of my setups.

Presuming one is already done with the level & distance setting, crossover settings, as well as aligning speakers and sub, one may find oneself with the ff. summed L+R response measured at the MLP:

1604127052262.png


Since there is only one sub in this bass managed configuration, we can apply EQ to that driver first:
1604127155896.png

output stage or after crossovers are applied (PEQ set 2 -- similar to miniDSP, JRiver can arrange and group one's set of correction filters in a certain order)

Placement and room acoustics situation between the left and rear right speakers (rears too) is not 100% symmetrical so I EQ each separately (ideal) in the bass area (as well as measure in SUM afterwards -- one may need to make EQ adjustments in sum or using shared PEQs):
1604127235148.png

(input stage or before crossovers are applied)
High above the transition zone, we can apply shared EQ correction for both.

Same is true for the rear channels in this 4.1 mch setup:
1604127303773.png

(input stage or before crossovers)

Single point measurements for each speaker are done as well as summed (L+R and fronts+rears) sweeps... but summing is much easier to do if you use the moving microphone method (MMM).

The LSR 305s (rears) HF response is already pretty flat on- and off-axis so there's not much to correct. The Sceptre S8 (coaxial main fronts), while having good beamwidth and directivity control isn't as straightforwardly "flat" or smooth and so requires a bit more care to EQ all the way up there.

Several measurements were necessary, both nearfield, at the main listening position, and everywhere in between! LOL I have more than one seating position (and I like to move around), so adjustments have to made along the way even if one can easily get a perfectly flat response right at the MLP.

With bass frequencies, I use Var smoothing. However, I prefer to use psychoacoustic smoothing above 1kHz first, then maybe 1/24 etc. only afterwards or at later stages.

Then I verify that my EQ adjustments are "relatively good" in as many ways as possible...

I recently re-did my EQ in that quite precarious, "speaker dominated" region... so might as well post some of the results from my adjustments here.

Very near-field (30cm) from speaker horn-throat log sweep; and separate MMM RTA+peak measurements (about half a meter away)
1604128097209.png


Correction works equally well for both left & right speakers
1604128129848.png


MMM right in front of MLP at 1.5m distance (straight on-axis)
1604128169786.png


At actual MLP 2m distance which is 15 degrees off-axis (exaggerated toe-in is applied)
1604128329645.png

Purple and blue are corrected responses

How does my EQ affect the off-axis response well beyond 15 degrees?
1604128433846.png


1604128443697.png

Looks like an improvement.

My vertical measurements below may not be so accurate, and I was getting tired so I stopped at 15 degrees...
1604128553697.png


1604128558365.png

We also see improvements here, too.


And finally, for the big picture (or summed response at all seating areas):
1604128762361.png


In this particular exercise, the sub channel was relatively easy to EQ as a single unit. BUT, the bass area around the transition zone was a little more difficult to balance across all couch seating areas and across all speakers -- hell, I don't always like to sit at the exact center sweet spot. The mids around the transition area, on the other hand, I decided best not to touch.

Lastly, even though I used a lot of PEQ bands (10) in the upper frequencies on my S8 main monitors, I 'corrected' or 'improved' just enough, and listened to esp. female spoken word audio tracks and music vocals to verify that the improvements were real as heard and not just imagined visually -- among a bunch of other test tracks. Having presets to quickly cycle between EQ settings while in the middle of playing test tracks helps tremendously here. I also deliberately tried to limit my use of high Q or narrow bandwidth filters. BTW, I didn't exactly apply every correction filter generated by REW -- manual adjustments and averaging were done and I didn't try to force-fit the FR to a given target specifically.

1604203399234.png


Above is an overlay of the 'averaged' Harman Target made by @thewas_ . It doesn't really fit precisely even with psychoacoustic smoothing in place, and that's totally fine! Ideally, one should not EQ by sight alone. One should also listen and make EQ adjustments by ear. In this setup, smooth and clear mids were my main priority as I do not use a center channel -- and this setup is primarily for movie watching. And while I can live with a little less upper bass, I felt that I needed more boost in the infrasonic region for those blockbuster movies and epic soundtracks, so I will be keeping it right there...

P.S. BTW, for my desk setup, my EQ settings are linked to as one i.e. it's all in mono. Measurements achieved there are way flatter and more symmetrical so it's perfectly okay.
 
Last edited:
OP
S

sfdoddsy

Active Member
Joined
Jun 6, 2019
Messages
293
Likes
438
Silly question, perhaps. If you are generating PEQ settings by measuring each speaker individually, does that mean you have a separate set of PEQs for each channel? Or do you combine the measurements somehow before generating the EQ filters?
That's the big question. Currently, post_EQ each of my speakers and sub is individually flat.

I can use the input EQ for MiniDSP or the linked output EQ mentioned above to apply the same settings to both. And I do this for speaker correction. Bbut those corrections are applied per speaker and I doubt each speaker will require the same EQ for room correction.

So you are making each speaker individually less accurate in the hope that the total response is more accurate.

Lots of trial and error seems to be the only way.

Assuming this worth doing in the first place.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
Why would anyone separate out the lower frequencies and re-mix it to both speakers in the recording away from the "effective location" of the rest of the directional harmonic content if that is likely to create phase difference between them. There might be some other technical reasons that I am not aware of but that looks like a false premise.

It's actually a standard practice for studio recordings. The reasoning goes something like this:
  1. Since we are unable to localise low-frequency sounds, there is no audible benefit to their being recorded/mixed in stereo in the first place.
  2. Since loudspeakers are most limited in terms of distortion/power handling at low frequencies, asking one speaker in a stereo system (i.e. no sub) to reproduce a greater share of the low-frequency content than the other speaker places an unnecessary (given our inability to localise these frequencies anyway) burden on that speaker.
  3. Standard practice in playback setups utilising a subwoofer is for the signal going to the sub(s) to be summed to mono before it is fed to the subwoofers. If this is done at the mixing/mastering stage, the engineer can ensure that this summing does not result in phase cancellation that would alter the low-frequency content of the mix, whereas there are no guarantees of this when it's done only at the reproduction stage by (e.g. a multichannel processor).
  4. In the case of vinyl masters specifically, record players are not capable of reproducing low-frequencies in stereo.
A number of authors have raised objections to points one and two, and some specialist reproduction systems using subwoofers indeed do not sum the sub channel to mono (see e.g. David Griesinger's work). However, this is certainly not a mainstream practice, and few recordings actually contain stereo bass content for the reasons listed above.

Having said that, you are much more likely to find stereo bass content on non-studio recordings of classical music (although I have no specific experience working on these, so I base that statement only on my general knowledge).
 
Last edited:

Chris A

Member
Joined
Jul 6, 2017
Messages
44
Likes
49
Location
Arlington, Texas
I think the authors forgot about humans having two ears instead of one, separated by an interaural distance between the eardrums (which varies from individual to individual). These human-to-human differences result in more than 90 degrees of phase shift at the eardrums vs. some frequency above 1 kHz. Where's the "correction" for these issues?

If one is going to try to set up their listening room for "head in a vise" reproduction (...I don't...), have a very high directivity (full range directivity at that) in their loudspeakers such that no significant room reflections occur, and they hear with only one ear (...I use two...), then that's great. I think the authors tried to make a complicated situation too simple. Einstein said that "everything should be made as simple as possible, but not simpler." That applies here most especially. I think the authors forgot how we hear. Unless you're using headphones and had typical-length ear canals, you're going to have issues with two loudspeaker channels separated more than 90 degrees of wavelength. That's why we have stereo, because of those phase differences that are introduced give the human hearing system cues that it doesn't get with monophonic reproduction.

Set your system boundary around the loudspeakers, room, listeners (having two ears...one ear doesn't do very much for stereo reproduction), AND the incoming music tracks--which themselves have massive amounts of phase distortion built-in (especially at low frequencies) unless they are direct-to-disc and have proven zero phase growth in the recording chain. I haven't ever seen nor heard of a canned recording with those capabilities, unfortunately.

I think you need to set up your loudspeakers/room transfer function (SPL and phase) to match that of either the mixing or mastering stations (I'd pick the mixing console room, and manually demaster the tracks back to downmix status to the degree that I could get them). The rest of this is really nonsense which usually comes from not applying a proper system boundary on the problem space, and considering the full problem space.

Chris
 

jlo

Active Member
Joined
Aug 3, 2018
Messages
100
Likes
177
This paper is very interesting and correlates to my own tests. I allways do both separated and correlated MMM measurements because at low frequencies, it is generally the L+R that counts. And when I calculate FIR correction, I use an hybrid correction with common low frequencies correction and separated high frequencies correction. Clearly in opposition to the standard way.
Hereunder is a comparison of the different methods :
- separated L and R correction, with incidence on L+R shown
- common correction based on L+R (mono) measurement
- hybrid correction



EQ.png
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
Having said that, you are much more likely to find stereo bass content on non-studio recordings of classical music (although I have no specific experience working on these, so I base that statement only on my general knowledge).

I believe I've come across such in "audiophile" recordings besides classical where the bass is clearly panned in the right or left. It can be a little disturbing and unexpected if you are not used to it.
 

andreasmaaan

Master Contributor
Forum Donor
Joined
Jun 19, 2018
Messages
6,652
Likes
9,399
I believe I've come across such in "audiophile" recordings besides classical where the bass is clearly panned in the right or left. It can be a little disturbing and unexpected if you are not used to it.

Yeh, it was quite common in the early days of stereo. Was it an older recording by any chance?
 
OP
S

sfdoddsy

Active Member
Joined
Jun 6, 2019
Messages
293
Likes
438
I've done a little test. Attached is a listening position measurement for my rear surround speakers. Measured individually, both exhibit a room related bump at around 200Hz.

I would normally EQ this.

Measured together (the blue line), the bump mostly disappears. I certainly wouldn't bother EQing it.
L R L+R.jpg
 
OP
S

sfdoddsy

Active Member
Joined
Jun 6, 2019
Messages
293
Likes
438
And this is all speakers measured together via the All Channel setting.

All speakers.jpg
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
Lots of trial and error seems to be the only way.

Assuming this worth doing in the first place.

YES! I would have stopped a long time ago, but playing around with audio related applications like REW every now and again can be kind of engrossing. Well, I presume a few people in here might find it to be the case. LOL

After several more rounds of experimentation, I think I may have found a better EQ 'speaker correction' that reaches down to 400Hz for my S8 monitors when used with an exaggerated toe-in...

Really!? Wait, is that really possible? Eh, well, I kind of took my cue from David Gunness himself:


1604453556614.png


Not exactly picture perfect like the graphic in the video, but an improvement nevertheless.
1604453562681.png


The speakers are about 15 degrees off-axis from the center MLP of my couch so that's what what I'm aiming for as can be seen above. Crossover is at 2.4 kHz, and according to the literature, maximum horizontal horn dispersion control is 110 degrees (x 90 degrees vertical) -- that's why I stopped at 55 deg.

Okay, so initially, I performed another round of MMM measurements at all seating positions and averaged them out. Created my correction EQ... then I added additional 'finer' PEQs by interactively referencing and re-adjusting the previous EQ settings using my own measured 'very near-field' (30cm!) directivity plots. I did not use a special target curve and just EQ'd it flat. Manual adjustments and a few extra PEQs post correction and re-measurements were necessary.

BTW, as should be obvious by now, completely flattening out the 15 degrees curve by itself (without referencing) makes the other off-axis curves a little worse-off.

1604454582849.png


While I was at it, I thought I might as well 'linearize' the overall bass more by adding two PEQs at around 100Hz and 200Hz.

1604454816336.png

Pretty much used up all the allocated 20 generic PEQ bands in REW now!

Log sweep at 2m Center MLP:
1604455677574.png


My rear channels aren't nearly as linear when measured at the center listening position...

1604455939805.gif


But that's because flattening it out completely makes the response at the far right and left seats even worse -- I don't want to do that.

So what about the sound? Well, it now sounds like the main monitors are playing exactly on-axis -- very focused and back to the original in-your-face vocals these monitors are well-known for -- voices are no longer "further back there" and diffuse sounding. Anyhows, while I was able to improve it before with just a 2dB boost around 2kHz, I believe this is a much better improvement overall.

Still, notwithstanding all the aforementioned improvement... can it be even further, further improved? Hmmmn, possibly, if Amir were to measure this monitor one might be able to create a more optimized, advanced convolution correction file for them. Dunno how much better it would be, though -- but my guess is only marginal improvements at best -- very likely nowhere near a night and day difference.
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
Additional directivity measurements from Resolution Magazine that I cleaned up a bit (just for the sake of comparison) below:

DIRECTIVITY 0 - Presonus Sceptre S8 - Resolution Magazine.png

A little different here and there

Graphs like these are very informative (beyond the basic on-axis FR) as they give you an idea how EQ corrections are going to affect the monitor's off-axis response.
 

StevenEleven

Addicted to Fun and Learning
Forum Donor
Joined
Dec 1, 2018
Messages
581
Likes
1,188
YES! I would have stopped a long time ago, but playing around with audio related applications like REW every now and again can be kind of engrossing. Well, I presume a few people in here might find it to be the case.

Still, notwithstanding all the aforementioned improvement... can it be even further, further improved? Hmmmn, possibly, if Amir were to measure this monitor one might be able to create a more optimized, advanced convolution correction file for them. Dunno how much better it would be, though -- but my guess is only marginal improvements at best -- very likely nowhere near a night and day improvement.

I have found the REW and trial and error tweak and remeasure this way and that way cycle quite engrossing. It is perhaps a trap set for the naturally curious. I’ve probably wasted three or four days of my life on it. But as a result, I can just put my system on full mono, coming out of five speakers and two subwoofers, and it’s relaxing and a nice tonal balance. I am free to move about the room and enjoy. I have an EQ setting just for that, I chop out the bass a little at 60 hz and below to keep it in good tonal balance. I kind of know what I’m doing because of all of the fiddling around and measuring in different configurations I’ve done with REW.

The unfortunate truth, I think, is that any system can be “even further further improved.” But I think you hit the nail on the head—I’ve come to feel that when the improvements stop being anything more than marginal, maybe it’s time to stop and enjoy using the gear for its intended purpose. It’s also very easy to for me start doing more harm than good at that point if I’m not careful.:)

So to answer the OPs question, in part, I guess, I keep the EQ for the front left and right speakers the same, I figure if I try to eq them differently from one another, my mind will pick up on nearly incomprehensibly complex visual and auditory cues and and it’d feel like something is not quite right. So in that sense I EQ in mono. I also have one configuration that is optimized for total mono sound from the whole 5.1 system. I also have variations on configurations for when I’m not using my subwoofers, for focusing on the tonal balance of my center speaker, for focusing on the tonal balance of my left and right speakers, etc.
 
Last edited:

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
3,106
Likes
2,313
Location
Canada
The unfortunate truth, I think, is that any system can be “even further further improved.” But I think you hit the nail on the head—I’ve come to feel that when the improvements stop being anything more than marginal, maybe it’s time to stop and enjoy using the gear for its intended purpose. It’s also very easy to for me start doing more harm than good at that point if I’m not careful.:)

I am reminded of these lines by Floyd Toole: "...some of the problem lies in our interpretations of measurements made in small rooms. The horrendously irregular steady-state “room curves” that we see simply do not correspond to what we hear. Did our problems begin when we started to make measurements? ... "

I was already quite happy with all of my speakers out of the box even without measurements. I believe it was after my purchase of a sub that actually got me into my measurements 'craze'... as big subs make room modes/standing wave peaks virtually impossible to ignore. But what is even more ridiculous is the fact that I frequently watch shows/movies on a portable smart phone or laptop out of convenience. I shouldn't be the type that's hard to please at all given my (clearly) substandard proclivities, but all those horrible wiggly lines in REW just looks like is it's begging to be 'corrected'. LOL
 
Top Bottom