I've messed around with this, and wasn't happy with the effect. By mixing in an inverted signal, dynamics suffer and bass suffers. (Because it's out-of-phase.)
A plain ol' hunk of styrofoam between the speakers works quite well. About as low tech as it gets, and ugly as sin, but it works. That's how I had the speakers set up in my office. I use a 42" TV for a computer monitor and a barrier between the speakers that goes nearly all the way to my face!
If that's too ugly, another thing that works nicely is to add a second set of speakers near the listening position, but delayed so that they're in phase with the stereo speakers. It basically 'augments' what's coming from the stereo speakers, and exaggerates the soundstaging cues.
This last one works shocking well (better than a barrier IMHO) and is dead-easy to implement with MiniDSP.
All the solutions mentioned above require you to know exactly where you will be seated, but for a system in an office or a small room, that's no big deal.
Patrick in my case I was going to try the inverted signal on the top end (8-10k+) and then compare the heils in phase for typical super tweeter top end reinforcement (trying with and without band passing main tweeter). Mostly experimentation, not trying to solve anything.
Augmentation is what I meant by low level satellites but at that time I had no dsp and had to rely on placement for correct delay, which ended up roughly 6’ high and maybe 2’ to sides and 1’ fwd of mains…..it really added to the soundstage but it had to be fine tuned to listening position as you stated…….I don’t dance around much anymore so not a big deal, and unless classicalfan breaks out the baton for channeling his inner Arthur Fiedler I’m thinking he probably listens at a fixed lp ?
Augmentation is what I meant by low level satellites but at that time I had no dsp and had to rely on placement for correct delay, which ended up roughly 6’ high and maybe 2’ to sides and 1’ fwd of mains…..it really added to the soundstage but it had to be fine tuned to listening position as you stated…….I don’t dance around much anymore so not a big deal, and unless classicalfan breaks out the baton for channeling his inner Arthur Fiedler I’m thinking he probably listens at a fixed lp ?
If anyone is interested in how to actually do the interaural crosstalk cancellation (it's not just phase inversion, and not 8-10k), I can go into it, but I've detrailed the thread enough otherwise.
im Pretty sure it falls under the premise of soundstage manipulation, I know what I’m trying is not ‘textbook’ but it also is not costing anything extra to try wiring the heil’s up in different ways…….plan is to take a lead from the the right mains tweeter over to the left side heil and vice versa also swappping polarity on the heil’s compared to mains tweeters and figuring out what the best xo might be maybe with minidsp 2x4hd or just a series cap.
Edit: I think the proper way includes a little delay from what I remember so the minidsp would prolly be the way to go.
Just something that’s in the queue to try out…..not gonna happen anytime soon as I’m way behind schedule on everything with higher priority!
I know, I know………what is higher priority than audio!? Tell that to the wife! 😢
Edit: I think the proper way includes a little delay from what I remember so the minidsp would prolly be the way to go.
Just something that’s in the queue to try out…..not gonna happen anytime soon as I’m way behind schedule on everything with higher priority!
I know, I know………what is higher priority than audio!? Tell that to the wife! 😢
Last edited:
I'm interested 🙂
Personally, i don't think the thread has been derailed by the discussion of what the "stereo" experience is capable of, and what it is not.
(I've encountered a number of the points you have made about stereo, while simultaneously reading Newell"s book, Recording Studio Design. particularly in Ch16, 'The Non-Environment Control Room.' )
Would also be interested in Michael Gerzon's three-speaker system, if you have any leads. Thx.
Personally, i don't think the thread has been derailed by the discussion of what the "stereo" experience is capable of, and what it is not.
(I've encountered a number of the points you have made about stereo, while simultaneously reading Newell"s book, Recording Studio Design. particularly in Ch16, 'The Non-Environment Control Room.' )
Would also be interested in Michael Gerzon's three-speaker system, if you have any leads. Thx.
I would be interested, jaddie. I have a pair of the SDA 1c Polks, but I haven't played with them for a while. On certain recordings they do have a large soundstage, but I much prefer the clarity and imaging of my horns. Glenn.
Yeah...well, its a start, but that won't work very well at all.
No, the WIFE is the highest priority! Just ask her.
Well spill some knowledge jaddie!
edit: isn’t failure supposed to be one of the best educators? I should be really really smart by now! 😆
Is it on topic enough classicalfan?
edit: isn’t failure supposed to be one of the best educators? I should be really really smart by now! 😆
Is it on topic enough classicalfan?
Last edited:
The basic idea is to attempt to cancel at least partially that arrives at the ear from the opposite speaker. So the right ear always hears the left speaker, and that's what needs to be at least reduced.
If you look at what happens to the sound from the left speaker as it arrives at the right, it will give you a clue. It's delayed slightly, depending on speaker angle from center (not distance from you), and it gets modified in spectrum by the head, chest, and pinna, but only from somehwhere mid-band and up, bass doesn't change much, and must remain in phase with the other speaker.
The response curve you have to add is somewhat bell shaped, with the peak around 2kHz, falling as frequency goes up and down. The time delay, actually ITD (Interaural Time Delay) is actually fairly complex, and many models have been attempted. However, you could generalize it to somewhere between 300uS and 700uS, depending on speaker angle. That's why early devices for this used all-pass filters, or bucket-brigade CCD delays. Today digital delay is fairly cheap. This entire thing could be achieved with a single device, probably something from Behringer or miniDSP, though I don't have specifics right now. The bell-curve is extremely approximate! The real HRTF is actually fairly complex, but speakers and the room will pretty much dominate anyway, so the general curve is a good place to start.
The cancellation signal will always be somewhat lower in SPL than the mains, so you need a way to control it.
And in all cases, the speakers should be in similar acoustic environments, free from reflections closer than 10' or so. Otherwise the actual crosstalk from speakers to ears is not symmetrical and impossible to cancel using identical processes.
So if you try to do this with a second pair of speakers, the basic function would be to invert polarity relative to the mains, apply a response bell-curve centered at 2kHz, with 6dB/octave slope to keep phase a minimum, then play with the position of the speaker to achieve a sub 700uS delay between the acoustic center of both the main and cancellation speaker. Since you have a 6dB roll-off downward in frequency, the "correction" will fade before it interacts with bass too much, as will the very high end.
This will give you a reasonable, though general, form of interaural crosstalk cancellation. You'll have to tune it by changing the response curve, delay, and level of the cancellation signal to fit your setup, but all part of the fun. And, unfortunately, to get this to work well, your head position must be precisely on the line perpendicutal to the midpoint of the line between the speakers.
For a more generally audible result, rather than feed just L>R and R>L, you can use a matrix to develp an L-R signal, the difference between Left and Right. You then apply the same filter, delay, and level control to that signal, and add it like this: L+(L-R) then R+(R-L). To get the R-L, you invert L-R. I did exactly this, but without the delay, in 1980. I had no cheap way to do a delay, and hated the performance of BBD devices. It actually works pretty well, even without the delay. Even made circuit boards, and sold a few. Kind of fun. And the wire-wrapped prototype still works today!
Hope that helps, or is at least interesting.
If you look at what happens to the sound from the left speaker as it arrives at the right, it will give you a clue. It's delayed slightly, depending on speaker angle from center (not distance from you), and it gets modified in spectrum by the head, chest, and pinna, but only from somehwhere mid-band and up, bass doesn't change much, and must remain in phase with the other speaker.
The response curve you have to add is somewhat bell shaped, with the peak around 2kHz, falling as frequency goes up and down. The time delay, actually ITD (Interaural Time Delay) is actually fairly complex, and many models have been attempted. However, you could generalize it to somewhere between 300uS and 700uS, depending on speaker angle. That's why early devices for this used all-pass filters, or bucket-brigade CCD delays. Today digital delay is fairly cheap. This entire thing could be achieved with a single device, probably something from Behringer or miniDSP, though I don't have specifics right now. The bell-curve is extremely approximate! The real HRTF is actually fairly complex, but speakers and the room will pretty much dominate anyway, so the general curve is a good place to start.
The cancellation signal will always be somewhat lower in SPL than the mains, so you need a way to control it.
And in all cases, the speakers should be in similar acoustic environments, free from reflections closer than 10' or so. Otherwise the actual crosstalk from speakers to ears is not symmetrical and impossible to cancel using identical processes.
So if you try to do this with a second pair of speakers, the basic function would be to invert polarity relative to the mains, apply a response bell-curve centered at 2kHz, with 6dB/octave slope to keep phase a minimum, then play with the position of the speaker to achieve a sub 700uS delay between the acoustic center of both the main and cancellation speaker. Since you have a 6dB roll-off downward in frequency, the "correction" will fade before it interacts with bass too much, as will the very high end.
This will give you a reasonable, though general, form of interaural crosstalk cancellation. You'll have to tune it by changing the response curve, delay, and level of the cancellation signal to fit your setup, but all part of the fun. And, unfortunately, to get this to work well, your head position must be precisely on the line perpendicutal to the midpoint of the line between the speakers.
For a more generally audible result, rather than feed just L>R and R>L, you can use a matrix to develp an L-R signal, the difference between Left and Right. You then apply the same filter, delay, and level control to that signal, and add it like this: L+(L-R) then R+(R-L). To get the R-L, you invert L-R. I did exactly this, but without the delay, in 1980. I had no cheap way to do a delay, and hated the performance of BBD devices. It actually works pretty well, even without the delay. Even made circuit boards, and sold a few. Kind of fun. And the wire-wrapped prototype still works today!
Hope that helps, or is at least interesting.
Elias has some nice three speaker stuff on his site, here matrix for Gerzon and some others http://elias.altervista.org/html/3_speaker_matrix.html
I never checked three speaker stereo sruff closely other than the single speaker cabinet stereo found in Elias site, which definitely has big sound stage (as I understand it) but not much precise imaging.
Thanks jaddie, I screen shot that and saved to my file for that project.
so bell curve peaking @ 2k 6db slope…….my heils can cross at 800 12db so that should be able to work the way you explain. 2x4hd should be able to handle all involved with the basics…..not too sure about the matrix setup.
actually 500uS is roughly 6” in distance so just setting the heil back that far from the baffle on the top of the mains should do it…….might just need to throw a series cap and inductor on the heils dialed @ 2k for the bell curve, wired l/r r/l and inverted polarity like I mentioned, level could be adjusted with amp gain (they have their own 60w x2).
Would that give a rough idea? Or is the matrix absolutely necessary?
so bell curve peaking @ 2k 6db slope…….my heils can cross at 800 12db so that should be able to work the way you explain. 2x4hd should be able to handle all involved with the basics…..not too sure about the matrix setup.
actually 500uS is roughly 6” in distance so just setting the heil back that far from the baffle on the top of the mains should do it…….might just need to throw a series cap and inductor on the heils dialed @ 2k for the bell curve, wired l/r r/l and inverted polarity like I mentioned, level could be adjusted with amp gain (they have their own 60w x2).
Would that give a rough idea? Or is the matrix absolutely necessary?
Last edited:
https://www.avsforum.com/threads/ho...-what-the-science-shows.3038828/post-57504496Maybe of interest to the OP'er classicalfan
snip
Allowing more room reflections can enhance the spatial presentation without destroying the timbre. Killing the room reflections puts you in a pristine direct sound field. Both sound timbrally neutral, differing only in spatial rendering. Because spatial effects are dominated by recording technique, it is nice to have alternatives.
This is not a new topic for me. 38 years ago I set up a listening room which had heavy velour drapes along the fronts of the side walls. When pulled out, the side wall reflections were absorbed, giving pinpoint, tight, imaging. When they were pushed back, they allowed for a more open, spacious soundstage. One tended to favor rock and pop, the other more spatially oriented classical recordings. Such an option only works well if one has well designed loudspeakers to begin with. /snip
snip
Allowing more room reflections can enhance the spatial presentation without destroying the timbre. Killing the room reflections puts you in a pristine direct sound field. Both sound timbrally neutral, differing only in spatial rendering. Because spatial effects are dominated by recording technique, it is nice to have alternatives.
This is not a new topic for me. 38 years ago I set up a listening room which had heavy velour drapes along the fronts of the side walls. When pulled out, the side wall reflections were absorbed, giving pinpoint, tight, imaging. When they were pushed back, they allowed for a more open, spacious soundstage. One tended to favor rock and pop, the other more spatially oriented classical recordings. Such an option only works well if one has well designed loudspeakers to begin with. /snip
Just a couple of cautions. 12dB/Oct crossover without the woofer will have a different phase response, and generally the composite of both drivers and the crossover is optimized to minimize this. If you just kill the woofer, will may not get the same phase response. Don't forget the top of the bell curve too.
Yes, the series cap would be a better way to go. Amp gain is your level control.
500uS is definitiely about 6", but the key is knowing where the radiation centers of the speakers are, the 0 time point, of both speakers. Normally you'd get that with measurement unless the additional speakers are identical models to the mains. But you have the right idea, there just might be some inaccuracy in the method. Getting the delay wrong will have an effect on how well this works because we're working with phase cancellation, and just a little difference in phase (time in this case) moves you away from the null pretty fast. So you can do the cut-and-try method of moving the new pair forward or back an inch at a time and see what happens. There should be a sweet spot. Use some highly separated material to test this.
The matrix is a different way to get the cancellation siganl, not technically more precise, in fact much more general beause it doesn't address the crosstalk that happens for center-panned instruments or sounds. They still present a crosstalk problem, and a resulting response null. What the matrix difference signal does do is create a more general correction, with some collateral ambience "recovery". It's another method, it will expand the soundstage too, and will be more clearly audible outside of the critical listening position.
If the matrix is beyond what you can do, go for the simple method first. It technically is more precise anyway.
The real problem with reflection is that the off-axis response of most speakers is neither smooth nor has a matched timbre to the on-axis response. In fact, it's the off-axis response that really affects how we perceive the coloration of any speaker, unless in a fully treated room. Speakers with smooth off-axis response, and smooth on-axis response sound better in untreated rooms because of this. Note the goal is "smooth" not "flat", which makes sense because "flat" response is generally not right even on-axis.
That's just wonderful! Isn't it fun to move the drapes and hear the change? I've done this with people too, they have a big blurry sound in a refelctive room where a solo voice sounds spread all over the room. Then I whip out a couple pieces of acoustic soak material, kill the worst reflection, and suddenly they can reach out and touch the singer. So much fun to demo!
This thread: https://www.diyaudio.com/community/threads/fixing-the-stereo-phantom-center.277519/ has quite a bit of info on the effects of cross talk and possible corrections. It also has measurements taken in a 'studio quality room' at the ear position to compare to.
https://www.diyaudio.com/community/...eo-phantom-center.277519/page-31#post-4746283
I've also done my own measurements and many experiments on this subject. It's fun and quite real when listening to Stereo. Especially when one diminishes the effect of early reflections this cross talk phenomenon becomes (much) more dominant. Speaker position and room layout influences the specifics of the delay.
In tests it becomes quite obvious if one finds the right delay though. It made a remarkable difference.
Personally I use a bit of mid/side EQ to 'cure' part of this problem, as well as using ambience speakers to help out further. I don't think the OP is willing to go that
far though. For Home Theater I still use something 'loosely inspired' by the cross talk cancellation schemes out there as well as some of Mr. Floyd Toole remarks
on the subject as I still use a phantom center.
https://www.diyaudio.com/community/...eo-phantom-center.277519/page-31#post-4746283
I've also done my own measurements and many experiments on this subject. It's fun and quite real when listening to Stereo. Especially when one diminishes the effect of early reflections this cross talk phenomenon becomes (much) more dominant. Speaker position and room layout influences the specifics of the delay.
In tests it becomes quite obvious if one finds the right delay though. It made a remarkable difference.
Personally I use a bit of mid/side EQ to 'cure' part of this problem, as well as using ambience speakers to help out further. I don't think the OP is willing to go that
far though. For Home Theater I still use something 'loosely inspired' by the cross talk cancellation schemes out there as well as some of Mr. Floyd Toole remarks
on the subject as I still use a phantom center.
Last edited by a moderator:
In addition, when setting gain this way the crossover frequency changes with level.
Um...I don't think so. The crossover is fixed at the speaker. The amp gain changes level only because it's at the amp input.
Three factors are linked.. the response, level and crossover frequency. All three curves shown here are LR2 @ 1kHz...
Allen, not if you're using one extra tweeter (per channel) for cross talk cancellation. It isn't in need of a complimenting woofer.
It's a separate bell shaped signal shaped around 2 Khz that can be changed by the amplifier gain to adjust the level of compensation.
Moving it backward/forward in position with regards to speaker placements adjusts it's delay.
It's a separate bell shaped signal shaped around 2 Khz that can be changed by the amplifier gain to adjust the level of compensation.
Moving it backward/forward in position with regards to speaker placements adjusts it's delay.