Yeah the situation is fine if I just move to the correct position and listen there but it is not practical as its a family livingroom and can't dominate it with stereo. Could be happy here but if the distance could be increased ever so slightly it would make sound at practical listening position a joy. Triangle is roughly 2.6m now and when I'm about meter in, so distance from ear to each speaker becomes roughly 2m the sound happens, not too far of, but still some.
Thanks for the tips, will do! I've only so far resorted to reasoning and had only few experiments for the basis so not much understanding on this yet. I bet looking at impulses could give more insight as you suggest.
Thanks for the tips, will do! I've only so far resorted to reasoning and had only few experiments for the basis so not much understanding on this yet. I bet looking at impulses could give more insight as you suggest.
I actually tried to see in REW and the phase response looks practially identical, the impulse is a bit shorter, but the same general shape, so it should work IMHO. Cutting down the WAW worked also getting almost identical results - but I would have to cut perfectly symmetrically I guess and I did not pay much attention to that.
An interesting problem, figuring it out will be helpful here. I wonder if, for the most part, it has to do with the angle of the speakers to your head? As the angle of each speaker approaches 90 degrees, you should feel more enveloped. Certainly that's been my experience, although all the way at 90 degrees wasn't ideal, maybe 70-80. Of course if you have strong reflections at the 2 meter position, that would influence the sound. How far apart are the speakers currently? Is the base of the triangle 2.6m?
Seeing some measurements at 2m and 2.6m could be enlightening, as @wesayso has mentioned. There are many things to look at.
I looked at the preset with 1024 taps. It fits perfectly at 1024, probably because nothing happens below 1kHz. Remember that left and right are different, you need both.
Yeah roughly so. At first I had the speakers in room corners, >5m apart, and phantom center was quite diffuse, had to get real close to baseline between speakers for some refinement to happen. Disappointed moved the speakers to the sidewall where is another possible placement for them. Roughly 3meters or so apart the transition moved further out fron the baseline. About 2m to each speaker. Mpved speakers closer together and the transition got bit more further room. Same ~2m distance to each speaker where the transition happens.
I thought could it be vertical reflections be part of it, as they would be roughly the same when distance from ear to speaker is same, but it could be multiple things like for example the angle you mebtion for enhanced envelopment. Thanks, something to experiment with.
Thanks for tips, I make measurements and post them here one day. Have you got suggestions what kind of listening experiments to conduct, to narrow things down?
Attached are two shuffle impulses made with rePhase. They are both stereo files at 44.1kHz sample rate, 1024 samples. One is 32 bit float, the other is in 24bit integer. Use whichever works for you. If you don't have rePhase or know how to make these at a different sample rate, just ask and I can post some here.
Quick test report, playing with toe-in seems to indicate I have multipart problem. When speakers are toed in 45deg, and I move forward the phantom center comes to focus pretty much where head is on-axis of the speaker as explained earlier. This seems to be mostly due to frequency response, if I point on-axis to listening position, less toe-in, some of the focus is now at listening position as well, victory. Waiting for waveguides which should have flatter DI on top octave and then readjust EQ, should enable the large-toe in and have time/intensity trading happening. Still missing the envelopment part at listening position, and might have to resort to some tricks like additional back / ambient channels like wesayso has been using, or something.
The success of that would rely on what kind of reflections you get in the first ~20 ms...
My level of reflections within that window is pretty low...
Hi Pano,
I tried the files right after your posted them, but had just begun an amplifier rack changeover, and wasn't sure I had all three LCR speakers totally correct.
Three 5-ways can be a mess...as the rear of my rack shows..(this mess is after the rework and full verification Lol)
Anyway, results before the amp changeovers were:
With 2-channel stereo and the normal file, both channels gave a solid center that was lower in tone than either left or right alone.
With 2-channel stereo and the phase shuffled file, both channels gave same solid center. The tonal difference between just left or right was definitely reduced.
With 3-channel LCR (and my fav so far, matrix setting), the normal file had a little bit more of the tonal difference removed, than the 2-ch phase shuffled.
With 3-ch LCR and the phase shuffled file, I really couldn't tell any difference between it and the normal file just described.
The amp rack changeover has had a bigger effect than i thought amps could make.
Before, I had a wild mix of different types/class amps driving different sections because of the evolving migration from 2 channel to 3-channel.
Now, QSC PL3s drive all three subs, CXDQ network amps drive all three lows, and all three mids.
The only difference in things now, is L&R coax CD's use Rotel hi-fi amps, and the center coax CD uses a CXDQ.
I have a third Rotel amp, but need to get another Q-Sys line out card to use it.
I'm also going to try to nab another CXDQ for the L&R CDs, just because I'm a little shocked at gaining clarity from swapping amps, and want to get things the same. (as well as compare the Rotels to the network amps)
Anyway, point of all that was to add my recommendation to what many have said,......... how important it is to have speakers (and apparently amps too) exactly the same....
Ok, Your files with the new amp set up.
2- ch stereo test didn't change much from before. Same lower tonality in phantor center, same degree of improvement with phase shuffling.
(previously, both left and right had matching amp setups; it was center than differed the most)
The 3-ch's tonality is now really close, left vs right vs center, with either the normal file or the phase-shuffled.
I'm really liking the LCR setup...it's like having the clarity and timing of mono, along with less blurred spaciousness.
Last edited:
Thanks so much for testing this a posting your results.
This is very interesting to me and I hope to others who read this thread. The phase shuffle does seem to have the desired tonal correction effect on your system. This is good to know. It’s also an important point that the shuffled file sounds nearly the same on the triphonic system. In theory a mono mix should sum correctly, but I have not heard it in practice.In your opinion, how close is the shuffled file in 2 channels mode to the normal file in triphonic mode? Does the shuffle do a decent job of imitating your center speaker? (At least as far as tonality.)
Since you did NOT say that you are going to throw out your center speaker
we can assume that the center speaker does something better than the shuffle, right? I hope that the test file has helped you achieve an even tonal balance across all three speakers.
Thanks go to you for the files, and all the work/initiative you've put into this thread !Thanks so much for testing this a posting your results.
With the LCR setup, it's difficult to say there was any tonality difference between L, R, and C. (with either the normal or shuffled file)In your opinion, how close is the shuffled file in 2 channels mode to the normal file in triphonic mode? Does the shuffle do a decent job of imitating your center speaker? (At least as far as tonality.)
Since you did NOT say that you are going to throw out your center speaker
In 2 channel stereo with the phase shuffled file, I could still hear some tonality difference between phantom C and L or R. (although clearly less than the normal file)
My gut call would be the shuffled file went about half as far as LCR, in terms of a center image having tonality that matched mono L or R.
Here's the LCR matrix that was being used.
Lout = (Lin-1.6dB) - (Rin-15.5dB)
Rout = corresponding
C = (Lin + Rin)-5.6dB
I'm still relatively new to the world of matrixing, but it appears to me this one isn't far from a simple sum L&R to center, that i think I've read you say gave pretty acceptable results (when you had a center speaker).
"Explain like I'm 5" - They say two adjacent drivers on a baffle working over the same frequency range will "comb" filter at some point, due to the center to center spacing.
Placing a mono speaker half way between the L / R speakers, well there's a big "center to center spacing", with a mono signal applied to all. How does that arrangement not turn into a comb filter - by the time the sound wavefront summation reaches the listener?
Consider Polk's "Stereo Dimensional Array"; where a driver arrangement is [(L+)<--5.5"-->(R-)] <--5'--> [(L-)<--5.5"-->(R+)] One would think with a mono signal applied, that would just comb the heck out of the sound too. Is this why "purists" dont make use of such tricks?
Placing a mono speaker half way between the L / R speakers, well there's a big "center to center spacing", with a mono signal applied to all. How does that arrangement not turn into a comb filter - by the time the sound wavefront summation reaches the listener?
Consider Polk's "Stereo Dimensional Array"; where a driver arrangement is [(L+)<--5.5"-->(R-)] <--5'--> [(L-)<--5.5"-->(R+)] One would think with a mono signal applied, that would just comb the heck out of the sound too. Is this why "purists" dont make use of such tricks?
There's a lot going on to try to distill it down to some thing really simple....beyond my pay grade
That said,
It's probably helpful to picture speakers as point sources with no comb filtering of their own, and visualize how their signals arrive in time and phase at the ears.
It's also helpful to remember that "stereo" itself is a trick...
... contrived through either multiple mic locations, multiple tracks summed together, studio manipulations, etc.
(So "purists" are just kidding themselves!)
It's a good thread...worth digging through if you have the time and inclination...
That said,
It's probably helpful to picture speakers as point sources with no comb filtering of their own, and visualize how their signals arrive in time and phase at the ears.
It's also helpful to remember that "stereo" itself is a trick...
... contrived through either multiple mic locations, multiple tracks summed together, studio manipulations, etc.
(So "purists" are just kidding themselves!)
It's a good thread...worth digging through if you have the time and inclination...
I'll try and entertain a smart 5 year old by mentioning what is happening and why. I'm pretty sure some people oppose these views but nothing I mention here can't be measured. Just take a dummy head and measure at one ear while both speakers get the exact same frequency sweep. Some folks are sure that Stereo is working "as intended" with this cross talk. After a lot of experimenting and trying different things/solutions I am way happier to have played with these effects as it has maximized my listening pleasure. So take from it what you want, I am just trying to share my point of view of what I prefer and why.
I guess it would depend on one's view of what's more pure (lol). The purist or a setup with a little help from one of the many mentioned solutions.
Here we go:
Your left ear hears the left speaker (with a mono signal). The right speaker will also be heard by the left ear (with that same mono signal playing) but that right sound will be delayed in time and with a reduced SPL level. The other (right) ear hears the right speaker, but also the left speaker, but that left sound will be delayed in time and reduced in level.
Both time delayed signals create combing when they combine with the direct sound. This combing discussed here will happen in every 2 channel Stereo setup.
An average room is full of reflective surfaces, like walls, ceiling and floor. While the speakers are playing, these surfaces get hit by the sound and reflect it towards the listener's ears with some delay. These reflections, delayed even further, will 'partially' fill the comb patterns we just discussed (and create new, more dense ones in the process). End result: you will hear Stereo and all things are fine in the world.
Now let's use a more directional speaker. Or place absorbing panels at first reflections. Or even use speakers that avoid certain reflections. Now we can actually hear these combing effects we discus in this thread, and notice the tonal differences between phantom center and the left speaker or phantom center and our right speaker. I wouldn't regard an OB speaker as more directional in this sense, as the back wave will reflect back to the listening spot at som,e point and act like that regular room, it will also help hide the discussed combing effects.
If you observe the picture in this post, you'd probably see that this problem we speak of won't happen at all frequencies. Our head provides shading for higher notes and at lower notes the wavelengths become so long that this combing does not happen at these frequencies any longer.
This thread is about that tonal difference, and suggests a few things we can do to counter the tonal differences. Like Pano's phase shuffler. If you have a regular room, no or not too many absorbing panels and speakers with an average wide coverage, like direct radiators, you might not be bugged about this tonal difference. Chances are that users of Horn speakers will notice this tonal difference sooner, if these horns have a tighter than average coverage pattern.
Even though that Polk arrangement had drivers next to each other that create combing, when viewed as a casual observer, what they were trying to do is actually lessen the comb filtering that happens at each ear. The succes of it largely depends on placement and room details. But it isn't a bad proposition if the user follows the strict guide lines that should come with such a setup. You may have read about using a mattress in front of your nose between the speakers to enhance the Stereo effect. What Polk meant to do is create the effect of that mattress, just like the Ambiophonic crowd is doing with software. As a side venture on the same team, Bacch was created by a former Ambiophonic research member (Prof. Edgar Choueiri), and is now being sold trough dealers. They all try to solve that same combing problem. Which really isn't a problem if you're in a room with speakers that interact with that room. Unless you want or have something different.
If it's not a problem you notice in your room with your speakers, you have no problem and need not solve anything. Don't buy the Polk speakers. They won't do you any good. If you do recognize this problem, read this thread a bit closer and maybe try a few of the tricks mentioned.
I guess it would depend on one's view of what's more pure (lol). The purist or a setup with a little help from one of the many mentioned solutions.
Here we go:
Your left ear hears the left speaker (with a mono signal). The right speaker will also be heard by the left ear (with that same mono signal playing) but that right sound will be delayed in time and with a reduced SPL level. The other (right) ear hears the right speaker, but also the left speaker, but that left sound will be delayed in time and reduced in level.
Both time delayed signals create combing when they combine with the direct sound. This combing discussed here will happen in every 2 channel Stereo setup.
An average room is full of reflective surfaces, like walls, ceiling and floor. While the speakers are playing, these surfaces get hit by the sound and reflect it towards the listener's ears with some delay. These reflections, delayed even further, will 'partially' fill the comb patterns we just discussed (and create new, more dense ones in the process). End result: you will hear Stereo and all things are fine in the world.
Now let's use a more directional speaker. Or place absorbing panels at first reflections. Or even use speakers that avoid certain reflections. Now we can actually hear these combing effects we discus in this thread, and notice the tonal differences between phantom center and the left speaker or phantom center and our right speaker. I wouldn't regard an OB speaker as more directional in this sense, as the back wave will reflect back to the listening spot at som,e point and act like that regular room, it will also help hide the discussed combing effects.
If you observe the picture in this post, you'd probably see that this problem we speak of won't happen at all frequencies. Our head provides shading for higher notes and at lower notes the wavelengths become so long that this combing does not happen at these frequencies any longer.
This thread is about that tonal difference, and suggests a few things we can do to counter the tonal differences. Like Pano's phase shuffler. If you have a regular room, no or not too many absorbing panels and speakers with an average wide coverage, like direct radiators, you might not be bugged about this tonal difference. Chances are that users of Horn speakers will notice this tonal difference sooner, if these horns have a tighter than average coverage pattern.
Even though that Polk arrangement had drivers next to each other that create combing, when viewed as a casual observer, what they were trying to do is actually lessen the comb filtering that happens at each ear. The succes of it largely depends on placement and room details. But it isn't a bad proposition if the user follows the strict guide lines that should come with such a setup. You may have read about using a mattress in front of your nose between the speakers to enhance the Stereo effect. What Polk meant to do is create the effect of that mattress, just like the Ambiophonic crowd is doing with software. As a side venture on the same team, Bacch was created by a former Ambiophonic research member (Prof. Edgar Choueiri), and is now being sold trough dealers. They all try to solve that same combing problem. Which really isn't a problem if you're in a room with speakers that interact with that room. Unless you want or have something different.
If it's not a problem you notice in your room with your speakers, you have no problem and need not solve anything. Don't buy the Polk speakers
. They won't do you any good. If you do recognize this problem, read this thread a bit closer and maybe try a few of the tricks mentioned.
The only thing worth mentioning: with discrete channels, like in a 5.1 multi-channel track the center gets a discrete signal, that no other channel receives. In a stereo matrix, the center sum always contains some left and right signal. Be it at a reduced level. We also don't have a pure left or pure right signal within the matrix solutions that goes to these specific positions. Something to keep in mind....
If the solution pleases you, no problem. It still is different from multi channel. Within songs mixed for Stereo there may still be differences in tonal balance even with this matrix setup due to how it was mixed originally.
Nice Wesayso.....glad you posted that...
I had written and deleted the below, instead opting for the simple message that i did post
Pls Do check out the italic paragraph though
Guys like Pano and Wesayso are way farther along with this than I, and will no doubt have better explanations....
But here's my middle-school try....
For any given signal, acoustic comb filtering is a function of both of driver/speaker distance spacing and time. Both must be considered together.
Take 2 perfectly matched point-source speakers outdoors (so we can also mainly ignore room reflections.)
Set them up in an equilateral/isosceles triangle to the listening position, where your head/or mic is perfectly centered between the speakers.
Play the same mono recording through both of the speakers.
Will you get comb filtering? At a centered measurement mic? At the ears?
My take is, if the measurement mic's capsule is sufficiently small and truly centered, the two speakers will sum coherently throughout the spectrum with no comb filtering.....both sides arrive in phase together.
The ears however, present a problem. Because each ear can hear both speakers, and ears are separated by distance.
I guess we could block one ear, and move the other ear to the center spot, to get rid of combing haha. But another way you sometimes see is to put a sound barrier between the speakers' center running out to the listening position, butting into the listeners head, so that only the left speaker hits the left ear, and right speaker hits the right ear. That should eliminate combing too, unless your mellon is like mine, empty on one side and dense on the other
Anyway, that L-R combing in perfect situations because of ear spacing, is a big part of the stereo combing dilemma. my 2c try..
I had written and deleted the below, instead opting for the simple message that i did post
Pls Do check out the italic paragraph though
Guys like Pano and Wesayso are way farther along with this than I, and will no doubt have better explanations....
But here's my middle-school try....
For any given signal, acoustic comb filtering is a function of both of driver/speaker distance spacing and time. Both must be considered together.
Take 2 perfectly matched point-source speakers outdoors (so we can also mainly ignore room reflections.)
Set them up in an equilateral/isosceles triangle to the listening position, where your head/or mic is perfectly centered between the speakers.
Play the same mono recording through both of the speakers.
Will you get comb filtering? At a centered measurement mic? At the ears?
My take is, if the measurement mic's capsule is sufficiently small and truly centered, the two speakers will sum coherently throughout the spectrum with no comb filtering.....both sides arrive in phase together.
The ears however, present a problem. Because each ear can hear both speakers, and ears are separated by distance.
I guess we could block one ear, and move the other ear to the center spot, to get rid of combing haha. But another way you sometimes see is to put a sound barrier between the speakers' center running out to the listening position, butting into the listeners head, so that only the left speaker hits the left ear, and right speaker hits the right ear. That should eliminate combing too, unless your mellon is like mine, empty on one side and dense on the other
Anyway, that L-R combing in perfect situations because of ear spacing, is a big part of the stereo combing dilemma. my 2c try..