The head shadowing effect is there wherever you place the speakers. It is a relative tonal problem induced by the comb filtering from ear distances. Equalisation can work but it has to be applied to the mid and side signals separately.
I think here is way to test it, just turn head 90 degrees so that path length from each speaker to each ear is the same and comb filter disappears. This makes vocals, snare drum, stuff on the phantom center sound brighter in general, don't worry about stereo image losing out just listen change in timbre of stuff that is center panned. What you all say would this simple test help identify and evaluate the "issue"?
Borrowing picture Wesayso posted:
Did some late night listening yesterday experimenting with head rotation and timbre change is relatively apparent with controlled directivity loudspeaker system I have, even though I have no idea what the magnitude of problem is for others and other systems. Sometimes it made for better sound on some songs, and on some not so much I think, while sound was "brighter" it felt more right/balanced with comb filter happening. Rotating head ever so slightly already starts to change the color of phantom center content and I think everyone can evaluate the influence of the comb filter simply by rotating head and listening critically timbre of center panned instruments on a musical piece. If there is no difference then room sound dominates direct sound, no comb-filter, no problem.
@mark100 as you have trinaural setup running and like to actively change balance with DSP, do you ever find your self tuning the center speaker balance only?
ps.
I found phantom center become brighter when listening very close to line between speakers but now the image got too wide, speakers in too great of an angle. Also noticed that when toe-in was adjusted and front wall had more attenuation to it, more off-axis, the phantom center lifted up in height indicating the sound was now at brains mercy and not localized to front wall / objects between speakers and stuff. Balancing out everything is a tricky business
Borrowing picture Wesayso posted:
Did some late night listening yesterday experimenting with head rotation and timbre change is relatively apparent with controlled directivity loudspeaker system I have, even though I have no idea what the magnitude of problem is for others and other systems. Sometimes it made for better sound on some songs, and on some not so much I think, while sound was "brighter" it felt more right/balanced with comb filter happening. Rotating head ever so slightly already starts to change the color of phantom center content and I think everyone can evaluate the influence of the comb filter simply by rotating head and listening critically timbre of center panned instruments on a musical piece. If there is no difference then room sound dominates direct sound, no comb-filter, no problem.
@mark100 as you have trinaural setup running and like to actively change balance with DSP, do you ever find your self tuning the center speaker balance only?
ps.
I found phantom center become brighter when listening very close to line between speakers but now the image got too wide, speakers in too great of an angle. Also noticed that when toe-in was adjusted and front wall had more attenuation to it, more off-axis, the phantom center lifted up in height indicating the sound was now at brains mercy and not localized to front wall / objects between speakers and stuff. Balancing out everything is a tricky business
Mono played back on two speakers vs the stereo version of that song on the same two speakers? I'm not surprised you didn't hear a (tonal) difference.
(I do believe I have that exact same album though, the 2003 RCA/BMG Heritage CD, pink disk)
The mono version is superior to the stereo version that is included on that album i.m.h.o. which seems more of an after thought, but the song features Grace in the center on both tracks, so she gets exactly the same cross talk, when listening to it in either mono or Stereo played back on the same speakers. The only difference perhaps is a slight SPL cut on the vocals of the stereo version.
Grace's voice is in the center in both versions you listened to and it would result in having the exact same peaks and dips if you listened to both versions on the exact same speakers. So what did you listen for? Did you focus on the background vocals alone? Because that would be the only thing that's different from a tonal balance point of view between mono and stereo. As those change from center to sides...
With your setup, adjusting toe in and position carefully (a mm at the time) so you get the best balance etc, you won't change the key factors of this stereo flaw, but it is quite possible you do adjust not the direct sound, but the level and timing of reflections just right so they help to hide this tonal problem.
For that to happen, if there are certain delayed but well timed reflections, differing slightly from each other so they create a finer comb pattern, they will help fill the cross talk nulls. They remain different enough in relation to the direct sound upon head movement too, with the key being that the pattern they create for the left and right ear have an ever so slight difference comb pattern, so they actually sum "right" at each ear. In fact, it's not unlike what the original shuffler in this thread was doing, just timed a bit later, but still arriving within the Haas limits.
It probably sounds more "full" or "complete" (for lack of better words) when you hit that key spot/adjustment. Not a marginal difference as it can make it sound better in more than just the sweet spot. Hard to explain perhaps but I've been there and I do think I know that difference you speak of, though I arrived there with other means as my speaker position is fixed. I'm actually working to get that back after a recent driver change in my speakers and a renovated living room, so all of the important properties are changed.
Here's what I mean:
The purple trace is what a theoretical cross talk pattern would look like for an averaged sized ear spacing in about a 3x3 m stereo triangle.
We see the nulls and an ideal sum of sound between the two slightly different timed direct sound sources at 3.7 ms 7.2 ms etc. (being +
6 dB in level)
The green trace is the ideal direct sound (of a single speaker), with an ideal reflection, timed just over 7 ms and 12 dB lower in level. We
don't live in an ideal situation like this, so we never get either that purple sum (that doesn't account for things like head shading etc. or
even head movement) or that green sum, as no reflection will be ideal over that broad a frequency range... but we need a bit of luck
(or time invested) to just get enough of a smaller comb pattern that hides/fills the holes and/or shaves off a bit of that +6 dB ideal sum.
As Disco Pete mentioned a 4K hole, that might actually be of some benefit, as the dip at ~1.8 Khz and the peak at ~3.7K (in this example,
yours may vary) probably are the most pronounced (aside from a balance difference between 200-800 Hz between phantom center and sides).
Move your head though, and almost all of it changes again. That's why some level of reflections is still welcome to hide this "problem", but we
could discus the ideal timing of it. The original shuffler is timed real early, but I've experienced that even a Haas kicker can still help hide the
stereo comb.
I shouldn't even call it a "problem". As most listeners in real rooms don't even realize it's there. As said before, it becomes noticeable in the
absence (or reduced/low level) of early reflections. Early reflections as in: all reflections still within the Haas limit.
The purple trace is what a theoretical cross talk pattern would look like for an averaged sized ear spacing in about a 3x3 m stereo triangle.
We see the nulls and an ideal sum of sound between the two slightly different timed direct sound sources at 3.7 ms 7.2 ms etc. (being +
6 dB in level)
The green trace is the ideal direct sound (of a single speaker), with an ideal reflection, timed just over 7 ms and 12 dB lower in level. We
don't live in an ideal situation like this, so we never get either that purple sum (that doesn't account for things like head shading etc. or
even head movement) or that green sum, as no reflection will be ideal over that broad a frequency range... but we need a bit of luck
(or time invested) to just get enough of a smaller comb pattern that hides/fills the holes and/or shaves off a bit of that +6 dB ideal sum.
As Disco Pete mentioned a 4K hole, that might actually be of some benefit, as the dip at ~1.8 Khz and the peak at ~3.7K (in this example,
yours may vary) probably are the most pronounced (aside from a balance difference between 200-800 Hz between phantom center and sides).
Move your head though, and almost all of it changes again. That's why some level of reflections is still welcome to hide this "problem", but we
could discus the ideal timing of it. The original shuffler is timed real early, but I've experienced that even a Haas kicker can still help hide the
stereo comb.
I shouldn't even call it a "problem". As most listeners in real rooms don't even realize it's there. As said before, it becomes noticeable in the
absence (or reduced/low level) of early reflections. Early reflections as in: all reflections still within the Haas limit.
Last edited:
Earlier in this thread there was the notion: wouldn't the mixing or mastering engineer not have fixed this problem? A lot of mixing most probably is done with a more extreme near field setup. Speakers up close on the desk... the combing will be less pronounced. Then there is the guarding of mono compatibility and headphone listening. Headphones don't suffer from this comb problem. So most of the time, there will still be a tonal difference between center and sides, but indeed it varies somewhat. There sure are differences between songs just like there are with regards to imaging. Some are awesome, some not so much. After playing with this for quite a while, the little adjustments I made in EQ and timing of reflections made most every song a little more pleasing to listen to. Not that many exceptions.
(listening to the exceptions trough headphones might identify the differences, if it sounds good on headphones, it can sound good with some compensation)
Though when watching movies with just a phantom center on my stereo this effect is way more 'exact' in results, as that center channel was mixed as center, not as a phantom center. So any and all help to adjust the balance will work each time. That's why I have a different way of compensating for stereo listening than for watching movies. The shuffler from the paper in the second post was meant for that problem. Using a phantom center for watching multi channel material. That sure works as advertised.
I'll shut up now...
The difference in mono vs stereo tonality that I've been talking about (that i think was the root of this thread) is 2ch stereo vs 1 channel mono (with L&R line level signals summed to mono).
So not 2-ch mono vs 2-ch stereo. It sounds like that's what you compared. Sorry if i'm misreading.
That 2-ch mono vs 2-ch stereo comparison often sounds nearly identical, surprisingly often. (I like to pop through the presets I can quickly compare.)
Sometimes i think 2-ch mono is a little more dynamic too, even when tonaility and image centering stay the same. Never know, track by track.
Also, if the mono version of the album is a simple summing of the stereo version L&R, then all is good for any comparisons. If not, seems all bets must be off.
Hi, all my tone and level controls are applied globally across all three LCR speakers equally.
What does change the center speaker's balance relative to L & R, are the matrix presets that allocate single among the L, R, & C.
That does entail a fine tuning, to the extent I wary the matrices available for comparisons.
I've relied on the very helpful work of Elias Pekonen, to build the matrices. http://elias.altervista.org/html/3_speaker_matrix.html
Don't mean to stray too far off topic, but there are 2 basic types of matrices as best i can tell: ones that use a simple linear constant to apportion between channels, and ones that use trig formulas to apportion in an energy preserving mode. (Elias does a great job explaining these, the works of Gerzon and others.)
Ok, leaving trinaural, and hopefully heading back fully on topic.
I'm questioning the range of frequencies that are combing, per the stereo depiction on the left. (thx Ron)
I've always thought the combing freqs would be in the middle some where, say 1-3kHz, where the head masks shorter wavelengths that can't wrap around the head, and longer wavelengths simply wouldn't comb as much due to less relative phase difference.
Two things have me scratching the mellon:
First, I tried Pano's normal and phase shuffled test files from #1064 as before, but this time cutting out driver sections one at a time.
When the lower two sections on the synergy were muted, and there was a steep high pass allowing response from only 750 Hz up, very little tonality difference could be heard (sides vs center, normal or shuffled)
Adding the mids, which span from 300-750Hz, defintely added tonality difference to the test.
As did adding the lows, 100-300Hz, which incrementally added more difference.
So Pano, now I'm hugely curious how the phase shuffled files are altered...
Second source of questioning what frequencies are in play is a quote from the link Daihedz gave in #1124 https://www.aktives-hoeren.de/viewtopic.php?f=40&t=13693....a quote from Uli Brueggemann, of Acourate.
5. Frequency Dependent Hearing
Gerzon correctly considers the frequency dependence. Above approx. 700 Hz, we hear essentially level-dependent (intensity stereophony), the ear distances are too large to be able to determine phase differences between the left and right ear signals.
Below 700 Hz we start to perceive phase differences (time-of-flight stereophony). At this point Gerzon then looks at the effects of rotation angle and then posits different angles for the respective areas. Which then leads to the fact that it makes sense to implement and coordinate their own equations for the respective areas. In reality, however, this also means that you have to separate frequency ranges. With conventional low-pass and high-pass filters, however, other phase shifts are introduced and, as is known, the addition or subtraction of phase-shifted signals is not necessarily effective. In this respect, Gerzon speaks of phase compensation. Today it is easier to use linear-phase crossovers, but there is a price to pay for a delay that is inherent to the principle.
Incidentally, Pekonen does not comment on this subject.
So anyway, always stuff to lean, huh?
The mono were the original recordings later remixed for stereo. But I'm not clear on 2ch mono vs 1ch mono? So for 1ch mono, I would just use one side of the pre and connect both ic with a Y connector for example? How does that make it different when the recording is two separate mono channels?
Hi, Yes, 1-ch mono is simply a single speaker playing both channels of a stereo track summed together.
I don't follow how a recording could be two separate mono channels? All regular recordings I've ever seen are either mono, or stereo.
It seems to me if an old mono recording like Grace Slick were converted to stereo, then the mono recording would need to be altered into two different distinct tracks, panning some consent L or R, altering some contents timing, etc etc, ...and then recorded as stereo.
If no alterations were done to the mono, and it is simply both the L & R stereo tracks, I would call that 2-ch mono.
I keep a 2-ch mono preset to test against stereo, just to see how much different it sounds. Many tracks are dang clos, meaning to me, that very little "stereo" mastering was actually applied. Hope that made sense.
So going back to the Jefferson Airplane Album.....
the test that i think this thread has been about, would be a stereo track, vs the same stereo track summed together and played over a single speaker.
Okay I’m clearly confused. If I’m playing a mono recording through left and right speakers, for example say an original mono Beatles album, what is coming out of my speakers?
Mono, seemingly coming from one speaker (phantom center!) between your stereo pair - if the speakers and room are really symmetrical and low distorsion. Otherwise, it will jump around and/or be a big(er) fat blob. If its really good, you have to sometimes peek to see if it's really 2 speakers. This is 2 speaker playing a mono track. Some swear by one speaker only for mono replay.
//
//
Like TNT says, mono is coming out of each speaker.
That's a good test for how well the speakers have matched response, and how well the room response is symmetrical.
There should be a very strong central image. If not, big problems (usually speaker, ime). And if an otherwise strong central image wanders some by frequency content, something is still amiss (usually more lack of room symmetry).
I'm also one of the guys who swears by one speaker mono. I think it's the way to measure and assess a speaker's listening performance....as it eliminates variables that are dang difficult to sort through ...like found in this thread !!!
And on some songs, they simply sound better to me one speaker mono, than any other way. They are usually very dynamic tracks, particularly dynamic bass.
Single speaker mono can also help improve truly sucky stereo recordings, imo.
Okay so again, as far as tonal imbalance goes, I hear no difference. In mono or in stereo I have fully integrated apparent sources. Both sound great but I do prefer stereo by a long shot.
Turning my head I definitely hear a change but it simply sounds as it seemingly would if I were listening to a live source, nothing abnormal.I think here is way to test it, just turn head 90 degrees so that path length from each speaker to each ear is the same and comb filter disappears. This makes vocals, snare drum, stuff on the phantom center sound brighter in general, don't worry about stereo image losing out just listen change in timbre of stuff that is center panned. What you all say would this simple test help identify and evaluate the "issue"?
Borrowing picture Wesayso posted:
View attachment 1089680 View attachment 1089679
Did some late night listening yesterday experimenting with head rotation and timbre change is relatively apparent with controlled directivity loudspeaker system I have, even though I have no idea what the magnitude of problem is for others and other systems. Sometimes it made for better sound on some songs, and on some not so much I think, while sound was "brighter" it felt more right/balanced with comb filter happening. Rotating head ever so slightly already starts to change the color of phantom center content and I think everyone can evaluate the influence of the comb filter simply by rotating head and listening critically timbre of center panned instruments on a musical piece. If there is no difference then room sound dominates direct sound, no comb-filter, no problem.
@mark100 as you have trinaural setup running and like to actively change balance with DSP, do you ever find your self tuning the center speaker balance only?
ps.
I found phantom center become brighter when listening very close to line between speakers but now the image got too wide, speakers in too great of an angle. Also noticed that when toe-in was adjusted and front wall had more attenuation to it, more off-axis, the phantom center lifted up in height indicating the sound was now at brains mercy and not localized to front wall / objects between speakers and stuff. Balancing out everything is a tricky business
This really is a subject for a different topic(or it's integral). I brought it up because I think critical placement takes care of this issue incidentally just because I find it swamps the issue of the op at best. The point being the actual stereo image intended by the engineer lives within that mm of tolerance. Outside of it you can not make an accurate judgement about this or any issue related to sound stage, period. I would wager if you can't confer, you have not heard a properly set up pair of speakers. And I'm claiming it holds true for dynamic speakers as well as planars/ESLs.
Must strongly disagree, from many long years of playing with this...on a bunch of speakers in a bunch of rooms.
I'm 100% sure i've heard properly set up pairs of speakers many times.....ESLs and dynamics.
besides, a stereo image 'intended by an engineer' exists only on the system he used to master with, given its spatial properties and the room it was in.
To think recapturing that is just a matter of finding the mm physical alignment on whatever system we are using, in whatever room, makes zero sense to me...