It is an issue of tonality not of sound stage. The link is that it becomes obvious when the same sound moves through the phantom centre or can be heard separately in different positions within the sound stage. I can only think that there is enough early reflections present in your setup to mask it given how much effort you have gone to in positioning.
It's not about recapturing what he heard, rather the effects he mixed into the recording.
I'm claiming any set of speakers requires this critical attention to optimize the entire system from the sound engineer's input to playback.
I know this is a controversial issue and I find it bewildering very few people have actually taken the time and effort to test this premise. It's easier to verify with planars because of their inherent behaviour, ie: beaming tendency. But because of this sharp focal characteristic you would think it's obvious, however there are other placement factors that have to be dealt with in order to verify it. Interestingly enough, because of the nature of dynamic drivers having a much more diffuse dispersion characteristic, it's even more difficult to pin point optimal placement, changes are more subtle. But there's no question when you hit the "G" spot. 😲
Last edited:
I'm not a fan of the idea we should try to recapture the recording engineers effects...i think that's more impossible than not, without having the same speakers in the same studio. etc.
But i surely agree with you in regards to taking the time and effort to optimize speaker placements (and room treatments).
Also agree that planars/electrostats with their unique focusing properties quickly teach one to focus on their focus
All of them that I've listened to, do have a magic setup where stereo comes alive. (when the stereo recording is good)
I think all this is off topic with the thread though. As i think even the planars/stats will exhibit the tonality shifts as have been discussed. (i know my X's do.)
I can’t argue with you except to say recapturing the engineer’s effects are part of the whole process. If you’ve done well, that will also be apparent insofar as imaging is concerned. This is what I was referring to.I'm not a fan of the idea we should try to recapture the recording engineers effects...i think that's more impossible than not, without having the same speakers in the same studio. etc.
But i surely agree with you in regards to taking the time and effort to optimize speaker placements (and room treatments).
Also agree that planars/electrostats with their unique focusing properties quickly teach one to focus on their focus
All of them that I've listened to, do have a magic setup where stereo comes alive. (when the stereo recording is good)
I think all this is off topic with the thread though. As i think even the planars/stats will exhibit the tonality shifts as have been discussed. (i know my X's do.)
Here is late night thought, what if decorrelating left and right was done with local acoustics rather than in the source? assume equilateral listening triangle and symmetric position in the room, good setup to make good phantom image and reveal the issue with head related comb filter. Now bring either speaker toward listening position some, 20cm for example, and then use delay and mild attenuation to this same channel.
What would happen, I think, is that the original direct sound(s) responsible of the phantom image would correlate again like they should but their first reflections from each side would not correlate, and perhaps help some with the combfilter. Or this not relevant at all, suppressed reflections contributing to localization is what makes the combfilter effect and this would be just one step back with system performance? Or have no effect at all perhaps. Anyone tried?
Or would it be better overall to have exactly symmetrical in room setup so that reflections from both speakers coincident exactly, perhaps making them both more like one? and have the combfilter
ps. Its interesting to think about reflections. For example the phantom image does not have reflections like real source on the center would. Instead, it has reflections worth of two sources, and whose physical location is not where the image is. Thinking sound panned to one side would have reflections of single source and where they should be. Comparing the two makes center always more diffuse, or unclear, with its presentation than a side panned sound. Unless, reflections that contribute to localization was minimized altogether.
What would happen, I think, is that the original direct sound(s) responsible of the phantom image would correlate again like they should but their first reflections from each side would not correlate, and perhaps help some with the combfilter. Or this not relevant at all, suppressed reflections contributing to localization is what makes the combfilter effect and this would be just one step back with system performance? Or have no effect at all perhaps. Anyone tried?
Or would it be better overall to have exactly symmetrical in room setup so that reflections from both speakers coincident exactly, perhaps making them both more like one? and have the combfilter
ps. Its interesting to think about reflections. For example the phantom image does not have reflections like real source on the center would. Instead, it has reflections worth of two sources, and whose physical location is not where the image is. Thinking sound panned to one side would have reflections of single source and where they should be. Comparing the two makes center always more diffuse, or unclear, with its presentation than a side panned sound. Unless, reflections that contribute to localization was minimized altogether.
Last edited:
Yes, this is a good point. A phantom center has a very different set of reflections than would an actual center speaker. IME that adds to the blur of the phantom image.
But I don’t think it will make much difference for late reflections, only early ones.
^ Yeah early ones I assume, those who have effect on localization. Localization cues in the source material would be intact but the local listening room cues would mix up some. I guess it would be similar situation as if the listening triangle was asymmetric to room, much easier to make happen and is probably what most people have already well, anyway, fun stuff.
Another setup to imagine:
Think about adding third speaker to a stereo system, real center source. One could bring it physically forward, closer to listener to increase path length difference to all room boundaries. Then use delay to bring the source sound (direct sound) back in time to the same distance as L and R speakers are. How would this be perceived? Direct sound would tell its where it should be but early reflections would be further in time, saying its closer. This, in my imagination, would make the center more clear than sides, already by having the center speaker of course. Anyway, fun stuff imagining
This train of though seems to lead to question how to make phantom image "reflections" equal to those of the real sources, the sides. Only way to do it with two speakers would be through minimizing the early (localization related) reflections to have phantom center as true as sides. Perhaps if reflections of each side are identical to each other, perfectly overlap for center panned material, and thus be like from one source instead of two. Which does not help with the head related comb filter though.. I guess I'm off topic but it seemed relevant thread to throw some thoughts at
well, anyway, fun stuff. Another setup to imagine:
Think about adding third speaker to a stereo system, real center source. One could bring it physically forward, closer to listener to increase path length difference to all room boundaries. Then use delay to bring the source sound (direct sound) back in time to the same distance as L and R speakers are. How would this be perceived? Direct sound would tell its where it should be but early reflections would be further in time, saying its closer. This, in my imagination, would make the center more clear than sides, already by having the center speaker of course. Anyway, fun stuff imagining
This train of though seems to lead to question how to make phantom image "reflections" equal to those of the real sources, the sides. Only way to do it with two speakers would be through minimizing the early (localization related) reflections to have phantom center as true as sides. Perhaps if reflections of each side are identical to each other, perfectly overlap for center panned material, and thus be like from one source instead of two. Which does not help with the head related comb filter though.. I guess I'm off topic but it seemed relevant thread to throw some thoughts at
Last edited:
Exactly, when the two speakers emit same sound so that they both arrive to ears simultaneously making brain think the sound came directly front. But, then comes first reflections that would, or not, reveal the sound source is not center but two sound sources left and right instead.
Thought was that we can manipulate the reflections by moving the speaker(s) along sides of equilateral listening triangle and then "correct" direct sound with delay to be like with equilater triangle setup.
But this is back to the topic, direct sound from each speaker arrives to both ears making the comb filter, which early reflections would usually fill out some. So the problem of the thread happens when early reflections are not loud enough to have effect, on timbre at least. Hense above thoughts are not good solution to the problem I think, early reflections. Perhaps if was relatively nearfield setup so that the 20cm was significant enough to scramble also the direct sound path lengths to the further ear around the face. I guess this would be too close though, <1m listening triangle.
edit. thinking more, the thread problem is due to head and two sources which makes combfilter, and the fix is usually early reflections that would fill the combfilter out. So there we have it, problem solved. But, the early reflections are bad for good stereo image. Two solutions to good image I outlined briefly was to have minimized early reflection, or perfectly matching early reflections. Logically we have now solved both problems when the early reflections are not killed but made perfectly matching instead, have early reflections and no combfilter, while still having good stereo image?
Last edited:
Don't forget that if the reflections of the phantom part don't line up, the other channel will actually help to lower the total effect of the reflection. It will average out and that isn't a bad thing.
The cross talk dips are helped by dissimilar reflection timing as well. But I wouldn't want them to arrive very early.
The cross talk dips are helped by dissimilar reflection timing as well. But I wouldn't want them to arrive very early.
^ True, different point of view, without localization mixed in so basically uneven reflections would make the phantom center have stronger direct sound than reflections, in comparison what sides have, as direct sound sums up from two sources but their reflections don't.
Would it be logical to think that the problem of the thread, comb filter, appears to perception when the room is too small and early reflections are too early so we must attenuate them, which results better image but cost is the combfilter?
So its problem only when the room is small, speakers relatively far out, and early reflection attenuation happening head sized wavelengths.
So solution would be to have bigger place so early reflections wouldn't need to be absorbed while still enjoying exceptional stereo image?
edit. basically I've just thought and wrote about something that describes the problem and not something that would solve it time for sleep
so basically uneven reflections would make the phantom center have stronger direct sound than reflections, in comparison what sides have, as direct sound sums up from two sources but their reflections don't. Would it be logical to think that the problem of the thread, comb filter, appears to perception when the room is too small and early reflections are too early so we must attenuate them, which results better image but cost is the combfilter?
So its problem only when the room is small, speakers relatively far out, and early reflection attenuation happening head sized wavelengths.
So solution would be to have bigger place so early reflections wouldn't need to be absorbed while still enjoying exceptional stereo image?
edit. basically I've just thought and wrote about something that describes the problem and not something that would solve it
time for sleep
Last edited:
Perhaps, I certainly wouldn't mind to have a listening room like Harman has: http://seanolive.blogspot.com/2009/05/harman-international-reference.html
Should I add my pay-pal address to start a fund raiser?
Should I add my pay-pal address to start a fund raiser?
Why not
But, from the above thought experiments there is more to be reasoned. When the comb filter appears to audibility it means the correlation between L and R is pretty good, not much early reflections fuzzing image, and everything is as goos as it gets stereo image vice. At least the system is capable of it, but perhaps the room is just tad too small, and/or listening triangle too big.
wesayso, do you have processing for this problem currently? I see you've made updates to the hardware and software.
But, from the above thought experiments there is more to be reasoned. When the comb filter appears to audibility it means the correlation between L and R is pretty good, not much early reflections fuzzing image, and everything is as goos as it gets stereo image vice. At least the system is capable of it, but perhaps the room is just tad too small, and/or listening triangle too big.
wesayso, do you have processing for this problem currently? I see you've made updates to the hardware and software.
Last edited:
Yeah they don't but the sides will, which intuitively breaks down the illusion, or at least affects it I think.
There is reflections when phantom center is happening, but its not the phantom that makes them but the two silent objects L and R to it as you say. Intuition says it would be confusing to brain and probably affects "clarity" of the phantom, or something. Must confuse brain some, do more processing than necessary.
Its brain (auditory system) that constructs the phantom center image from two direct sounds. My suspicion was that if reflections from both L and R speaker would overlap perfectly they would also make "single" set of reflections. Even though they come from different directions but like the phantom center brain could construct them to the center as well. If you think about it, they are bit shorter through sidewalls, longer through front wall, compared to if the phantom had real reflections. But all this would do is have "apparent" room size different than for side signals. However, attenuating/delaying all early reflections (enough) reflections that happen when phantom image appears would closely match those when the real sources (side panned) play and make more consistent total image. Third speaker in the center would fix it all.
Anyway, its just me speculating, imagining early reflections of actual center source and comparing them to two sources that make phantom center.
There is reflections when phantom center is happening, but its not the phantom that makes them but the two silent objects L and R to it as you say. Intuition says it would be confusing to brain and probably affects "clarity" of the phantom, or something. Must confuse brain some, do more processing than necessary.
Its brain (auditory system) that constructs the phantom center image from two direct sounds. My suspicion was that if reflections from both L and R speaker would overlap perfectly they would also make "single" set of reflections. Even though they come from different directions but like the phantom center brain could construct them to the center as well. If you think about it, they are bit shorter through sidewalls, longer through front wall, compared to if the phantom had real reflections. But all this would do is have "apparent" room size different than for side signals. However, attenuating/delaying all early reflections (enough) reflections that happen when phantom image appears would closely match those when the real sources (side panned) play and make more consistent total image. Third speaker in the center would fix it all.
Anyway, its just me speculating, imagining early reflections of actual center source and comparing them to two sources that make phantom center.
Always, I've tried every thing I could think of, but the one thing that worked best so far has been a little mid/side EQ for tonal balance.
I found that in my setup the phase shuffler had a slight SPL difference between channels, favoring one side. Cross talk cancellation was promising but it gave me listeners fatigue. Next to the mid/side EQ, which solves the tonal balance quite well, the ambience speakers I've got fill in most of the 'missing' info. I could live with this forever, but in the interest of learning something new I'll probably keep experimenting.
A shuffler like solution works very well for movies. With the discrete center channel it is much easier to solve the tonal variation. For music I still prefer the mid/side EQ + ambient channels, as it gives more of a 'holographic like' imaging.
I guess we need to set up stereo outdoors, and compare tonality of the phantom center to a side running in mono, to get a handle of the role of reflections.
Without doing so, my gut thought (which is worthless), is that reflections don't matter too much, assuming they are down enough in amplitude vs direct sound.
Also, must admit I'm really beginning to doubt the whole idea of comb filtering of frequencies that wrap around the head in a way they comb.....to produce the tonality shift as been discussed.
I'm still doing a boatload of comparative listening 2 ch stereo, vs 3 ch matrixed LCR, vs mono of any one speaker, vs 2 ch mono, etc etc....
How a track/album was mastered for stereo makes quite a difference in the relative audible tonality change between those comparisons.
That's not the big reason for doubting the head wrap stuff though.....
......doubting comes from plugging one ear up and making the same comparisons.
Then, i don't hear much difference in the mid-high and up frequencies, but low-mid and bass change rather dramatically, getting much fuller with 2 channel that with mono.
I dunno, seems two ears vs one has more to do with low end hearing/perception, than with mid & high. (Maybe time for a hearing check , yikes lol)
Without doing so, my gut thought (which is worthless
), is that reflections don't matter too much, assuming they are down enough in amplitude vs direct sound.Also, must admit I'm really beginning to doubt the whole idea of comb filtering of frequencies that wrap around the head in a way they comb.....to produce the tonality shift as been discussed.
I'm still doing a boatload of comparative listening 2 ch stereo, vs 3 ch matrixed LCR, vs mono of any one speaker, vs 2 ch mono, etc etc....
How a track/album was mastered for stereo makes quite a difference in the relative audible tonality change between those comparisons.
That's not the big reason for doubting the head wrap stuff though.....
......doubting comes from plugging one ear up and making the same comparisons.
Then, i don't hear much difference in the mid-high and up frequencies, but low-mid and bass change rather dramatically, getting much fuller with 2 channel that with mono.
I dunno, seems two ears vs one has more to do with low end hearing/perception, than with mid & high. (Maybe time for a hearing check , yikes lol)
You're not wrong about that, there is a big difference in bass perception of a single speaker mono vs stereo phantom.
That's part of the tonal balance difference of the phantom center and the sides this thread is about... Somewhere in the beginning of this thread there was an article linked about an S-curve, which is a form of mid/side EQ that was meant as a fix of that part of the problem. Sadly though, those pages that were linked are long gone. I did find a paper that was linked to, explaning the S curve idea (sorry, in german):
http://www.sengpielaudio.com/FrequenzabhHoerereignisrichtung.pdf
They speak of 400 and 4000 Hz, in my experience and my room I'd pick 650 Hz and my top end deviates way more from that graph, just a low Q drop at 4000 Hz for the sides would be a start though, while having no change to the phantom top end.
Toole shows the difference in a measurement between a phantom center vs a real center in his book. That should make the problem pretty obvious. That's posted here: https://www.diyaudio.com/community/...eo-phantom-center.277519/page-14#post-4693335
Member jim1961 showed measurements taken at ear position in this thread, I've done similar measurements myself using a foam ball as a dummy head.
It wouldn't be interesting to use any fine EQ above say 1 KHz if one moves around in the room. Only in the sweet spot, exactly equal distance to both speakers do the dips and peaks line up. The shuffler mentioned in this thread changes that symmetry, that's it's biggest accomplishment. Due to that change, the sweet spot actually sounds more clear, and tonal balance will improve too. If your room has a lower level of reflections to begin with. Here's a recent try from a member on ASR of the 'phase only' shuffler. It prevents the dips and peaks from the cross talk from becoming obvious. In a room with regular reflections you're not going to notice it to the same extend. just look at dips and peaks that originate from those reflections to find out the "why". The reflections, even at a lower level, do help fill in the blanks for our brain and as long as you're not at a symmetric spot, it will level out mostly anyway. But it might still have some of that tonal balance difference, you be the judge of that...
That's part of the tonal balance difference of the phantom center and the sides this thread is about... Somewhere in the beginning of this thread there was an article linked about an S-curve, which is a form of mid/side EQ that was meant as a fix of that part of the problem. Sadly though, those pages that were linked are long gone. I did find a paper that was linked to, explaning the S curve idea (sorry, in german):
http://www.sengpielaudio.com/FrequenzabhHoerereignisrichtung.pdf
They speak of 400 and 4000 Hz, in my experience and my room I'd pick 650 Hz and my top end deviates way more from that graph, just a low Q drop at 4000 Hz for the sides would be a start though, while having no change to the phantom top end.
Toole shows the difference in a measurement between a phantom center vs a real center in his book. That should make the problem pretty obvious. That's posted here: https://www.diyaudio.com/community/...eo-phantom-center.277519/page-14#post-4693335
Member jim1961 showed measurements taken at ear position in this thread, I've done similar measurements myself using a foam ball as a dummy head.
It wouldn't be interesting to use any fine EQ above say 1 KHz if one moves around in the room. Only in the sweet spot, exactly equal distance to both speakers do the dips and peaks line up. The shuffler mentioned in this thread changes that symmetry, that's it's biggest accomplishment. Due to that change, the sweet spot actually sounds more clear, and tonal balance will improve too. If your room has a lower level of reflections to begin with. Here's a recent try from a member on ASR of the 'phase only' shuffler. It prevents the dips and peaks from the cross talk from becoming obvious. In a room with regular reflections you're not going to notice it to the same extend. just look at dips and peaks that originate from those reflections to find out the "why". The reflections, even at a lower level, do help fill in the blanks for our brain and as long as you're not at a symmetric spot, it will level out mostly anyway. But it might still have some of that tonal balance difference, you be the judge of that...
I have set up many, many systems outside and in large venues where early reflections were tiny or absent. The dull phantom center is much more of a problem there than in small spaces. The effect is quite pronounced and is detrimental to dialog and other vocals. That is why since 1991 I mix with a center array or cluster. Vocals go there and it makes a huge difference in intelligibility.
The comb filtering is real and can be a significant problem. But in small live rooms with wide dispersion speakers, many people will never hear it.
The comb filtering is real and can be a significant problem. But in small live rooms with wide dispersion speakers, many people will never hear it.
Hi guys, i did a bad job explaining what i meant.
I did not mean to be saying i doubt that a phantom center has a downward tonality shift vs mono. I strongly know/experience that downward shift to be true.
In fact, the LCR matrix I'm using with a true center speaker has added so much enjoyment by adding the presence of great mono into the middle of the stereo field, I don't try to listen outdoors so much anymore.
Most (but not all) stereo without the center matrix, sounds kind of hollow and relatively muted in comparison to the LCR. So again, I'm a believer !
My doubts are about the cause of the change in tonality....and again all due to plugging up one ear, and comparing.
Then, the low end changes so much compared to the high end. It boosts with stereo, and I know a boost in low end is often perceived as a drop in the high end. That's all.
Also should say, level matching the various comparisons is tougher than i'd expect...and mismatches may be part of my puzzling.
Wesayso, thanks for the links....good refence stuff.
Pano, sounds like you did quite a bit of mixing FOH. Nice !
I did not mean to be saying i doubt that a phantom center has a downward tonality shift vs mono. I strongly know/experience that downward shift to be true.
In fact, the LCR matrix I'm using with a true center speaker has added so much enjoyment by adding the presence of great mono into the middle of the stereo field, I don't try to listen outdoors so much anymore.
Most (but not all) stereo without the center matrix, sounds kind of hollow and relatively muted in comparison to the LCR. So again, I'm a believer !
My doubts are about the cause of the change in tonality....and again all due to plugging up one ear, and comparing.
Then, the low end changes so much compared to the high end. It boosts with stereo, and I know a boost in low end is often perceived as a drop in the high end. That's all.
Also should say, level matching the various comparisons is tougher than i'd expect...and mismatches may be part of my puzzling.
Wesayso, thanks for the links....good refence stuff.
Pano, sounds like you did quite a bit of mixing FOH. Nice !