I was recently reading through some old threads regarding using three (identical) loudspeakers for stereo listening. The most heated debate was already years ago.
I would be really glad to hear from everyone, who is using a three loudspeaker setup for stereo.
My own experiment with the Single Speaker Stereo (three fullrangers matrixed in a single box) were quite encouraging and I would like to experiment a bit more - especially with extended bass and more SPL - I will share my experiments as they will come.
I am especially interested in configurations where the speakers are relatively close together in a straight line close to the front wall - as that is my physical limitation for speaker placement.
I would be really glad to hear from everyone, who is using a three loudspeaker setup for stereo.
My own experiment with the Single Speaker Stereo (three fullrangers matrixed in a single box) were quite encouraging and I would like to experiment a bit more - especially with extended bass and more SPL - I will share my experiments as they will come.
I am especially interested in configurations where the speakers are relatively close together in a straight line close to the front wall - as that is my physical limitation for speaker placement.
I do not (currently) use matrixed LCR, but according to Dolby Laboratories, it would reduce channel separation. That is, some content from the left & right would leak into the centre during matrixing, effectively reducing the width of the sound stage. Auto-balance / gain steering techniques (e.g in ProLogic) alleviate this problem, but only to an extent.
However, in discrete multi-channel mixes, the centre being a separate channel with its own content, doesn't disturb the interaction between the left/right pair.
However, in discrete multi-channel mixes, the centre being a separate channel with its own content, doesn't disturb the interaction between the left/right pair.
Last edited:
Djk was impressed by a mono (mixed stereo L+R) as the center.
I always thought close speakers separated by a 4" thick foam wall (ambiophonics) was best.
But that is a 2 speaker setup, not what you asked for.
2 speakers close together, the carver sonic holigraphy.
But the opamps used then aren't the greatest, and every cap would need to be replaced (age).
But again, not 3 speakers either.
I always thought close speakers separated by a 4" thick foam wall (ambiophonics) was best.
But that is a 2 speaker setup, not what you asked for.
2 speakers close together, the carver sonic holigraphy.
But the opamps used then aren't the greatest, and every cap would need to be replaced (age).
But again, not 3 speakers either.
Ok, some links. Lots of great information by Elias (unfortunately not active here anymore) http://elias.altervista.org/index.html
https://www.diyaudio.com/community/...equations-for-3-speaker-stereo-matrix.222881/
https://www.diyaudio.com/community/threads/making-3-front-channels-out-of-stereo-signal.193973/
Also inspired by @mark100 's 3 speaker setup https://www.diyaudio.com/community/threads/syn-10.383607/
https://www.diyaudio.com/community/...equations-for-3-speaker-stereo-matrix.222881/
https://www.diyaudio.com/community/threads/making-3-front-channels-out-of-stereo-signal.193973/
Also inspired by @mark100 's 3 speaker setup https://www.diyaudio.com/community/threads/syn-10.383607/
Some more links.
https://www.diyaudio.com/community/...ns-on-matrixed-single-stereo-speakers.257110/
This is what I based my experiment with SSS upon.
https://www.diyaudio.com/community/threads/stereophonic-sound-from-a-single-loudspeaker.200040/
And here are my 3FE22 speakers (SSS, Hydra clone and F1 clone) I use for my experiments. https://www.diyaudio.com/community/...-full-range-driver.379277/page-3#post-6864437
Especially the SSS with a mono subwoofer is super cool, but low in SPL.
https://www.diyaudio.com/community/...ns-on-matrixed-single-stereo-speakers.257110/
This is what I based my experiment with SSS upon.
https://www.diyaudio.com/community/threads/stereophonic-sound-from-a-single-loudspeaker.200040/
And here are my 3FE22 speakers (SSS, Hydra clone and F1 clone) I use for my experiments. https://www.diyaudio.com/community/...-full-range-driver.379277/page-3#post-6864437
Especially the SSS with a mono subwoofer is super cool, but low in SPL.
I am using a 3rd speaker placed back in the listening room that has its own amplifier and gets its signal from a substraction of L and R in a differential amplifier and amplifying the result. It has its own volume control and final amp.
The listening is much better than common stereo and when good signals are feed, it is very enjoyable. With low quality audio signals (MP3 @128KBS) the listening is unpleasant because too many artifacts. But at better sources (CD, Vynil, stereo FM, AAC3 from phones, MP3@320K) is very interesting mainly in live recordings. You listen really like being in the middle of the recording place. But increasing too much the volume of the 3rd channel makes listening un-natural. So the adjust of it is relatively critical. I made it several years ago and still in use.
The listening is much better than common stereo and when good signals are feed, it is very enjoyable. With low quality audio signals (MP3 @128KBS) the listening is unpleasant because too many artifacts. But at better sources (CD, Vynil, stereo FM, AAC3 from phones, MP3@320K) is very interesting mainly in live recordings. You listen really like being in the middle of the recording place. But increasing too much the volume of the 3rd channel makes listening un-natural. So the adjust of it is relatively critical. I made it several years ago and still in use.
Happy Holidays all !
Hi pelanj, looks like you've been doing some real homework on LCR matrix.
Lemme first say LCR matrix is giving me the first sound that is letting me forget about how much better sound is outdoors.
I love stereo's indoor effects......imaging and envelopment, but its lack of clarity compared to outdoor mono has been a real bummer for me.
Anyway, LCR matrix is solving the clarity tradeoff (albeit at a small, but totally worth it, expense to envelopment imo...no free lunch in audio, huh?)
Since I'm continuing to learn how to improve LCR matrix, maybe it's best to just say what I've found doesn't work so well.....
First thing on the list, is a different center speaker than L & R. Number 1 no-no, ime.
Needs to be identical, even including a subwoofer if subs are used on L&R.
2nd for me, has been the total listening angle encompassed. Less than 60 degrees doesn't differentiate LCR enough from stereo, and greater than 90 degrees looses the sense of a connected sound stage, far L to far R. Could just be my room though.
No delay on center, if all three are in a line..
Matrix needs the ability to delay center speaker to be same distance from listener as L&R.
Levels not equal on all three. I use pink noise to set equal SPLs at LP. ( this is where not having the same center speaker setup, shows how different pink sounds at an equal SPL to L&R)
Various forms of simpler matrices built off of L-R, R-L, R+L... have not worked as well as i've hoped....
or rather i should say they haven't wowed me like various forms the energy preservation matrices, ala Gerzon, etc,
By various forms, I mean different constant weighting coefficients for the simple guys, and different trig based coefficients (based on listening angle) for the energy preservation guys.
Lot to learn still, here. Want to try frequency dependent matrices still.
Apart from how a particular LCR matrix sounds in general, my favorite way to judge success, is what percent of stereo recordings does it obviously improve. Some stereo recordings are as good as it gets, but surprisingly few. Most recordings simply gain clarity, and a more solid center image with LCR.
I have two levels of center matrixing, one softer than the other for when a recording gets pulled too much to center and looses L & R spaciousness.
Which happened a lot, until my center speaker also included a sub. Then, the more aggressive matrix really started to shine, and is my current favorite.
It's why i think having the same center speaker is a must.
Anyway, the point about what type matrix works or doesn't work, is about the need to be able to try stuff easily....and verify that the setup is working as expected.
So the last 'what doesn't work', is not having a way to quickly check the sound from each speaker independently, for a good smell test.
And of course, being able to A/B stereo vs matrix.
Hi pelanj, looks like you've been doing some real homework on LCR matrix.
Lemme first say LCR matrix is giving me the first sound that is letting me forget about how much better sound is outdoors.
I love stereo's indoor effects......imaging and envelopment, but its lack of clarity compared to outdoor mono has been a real bummer for me.
Anyway, LCR matrix is solving the clarity tradeoff (albeit at a small, but totally worth it, expense to envelopment imo...no free lunch in audio, huh?)
Since I'm continuing to learn how to improve LCR matrix, maybe it's best to just say what I've found doesn't work so well.....
First thing on the list, is a different center speaker than L & R. Number 1 no-no, ime.
Needs to be identical, even including a subwoofer if subs are used on L&R.
2nd for me, has been the total listening angle encompassed. Less than 60 degrees doesn't differentiate LCR enough from stereo, and greater than 90 degrees looses the sense of a connected sound stage, far L to far R. Could just be my room though.
No delay on center, if all three are in a line..
Matrix needs the ability to delay center speaker to be same distance from listener as L&R.
Levels not equal on all three. I use pink noise to set equal SPLs at LP. ( this is where not having the same center speaker setup, shows how different pink sounds at an equal SPL to L&R)
Various forms of simpler matrices built off of L-R, R-L, R+L... have not worked as well as i've hoped....
or rather i should say they haven't wowed me like various forms the energy preservation matrices, ala Gerzon, etc,
By various forms, I mean different constant weighting coefficients for the simple guys, and different trig based coefficients (based on listening angle) for the energy preservation guys.
Lot to learn still, here. Want to try frequency dependent matrices still.
Apart from how a particular LCR matrix sounds in general, my favorite way to judge success, is what percent of stereo recordings does it obviously improve. Some stereo recordings are as good as it gets, but surprisingly few. Most recordings simply gain clarity, and a more solid center image with LCR.
I have two levels of center matrixing, one softer than the other for when a recording gets pulled too much to center and looses L & R spaciousness.
Which happened a lot, until my center speaker also included a sub. Then, the more aggressive matrix really started to shine, and is my current favorite.
It's why i think having the same center speaker is a must.
Anyway, the point about what type matrix works or doesn't work, is about the need to be able to try stuff easily....and verify that the setup is working as expected.
So the last 'what doesn't work', is not having a way to quickly check the sound from each speaker independently, for a good smell test.
And of course, being able to A/B stereo vs matrix.
Me too! I used to tell my dialog impaired friends to use their stereo receiver's summed mono mode.
You should really think again of the speaker placement and exploit the long wall in the room. Maybe you think it is to narrow and too wide but this is exactly how it should be made in order to be really good - much wider than you think is possible - but the centre will save you and the stage will be wonderful if you succeed with the right processing. But I could see that directivity might hinder you....
//
//
If you're long wall, then an L+R centre might help "pull" the vocals / bass to where they belong, especially if the angle subtended by L/R at the LP is large, say 90* or 120* etc. But the long-wall approach might also encourage stronger back-wall reflections, as the room is now shallower than before.
Alternatively, you may try this mid/side matrix:
https://www.musicdsp.org/en/latest/...ntrol-obtained-via-transfromation-matrix.html
https://www.musicdsp.org/en/latest/...ntrol-obtained-via-transfromation-matrix.html
I tried a mid/side matrix in winamp and it seems to emulate very well (for free) what the DPLII's centre-width function is said to do, that is, remove a weighted portion of the centre material from the L/R and restore the same via the centre.
The result of the matrix is:
Lout = L - slider*R
Cout = (L + R) * slider
Rout = R - slider*L
When played back through 3 identical speakers, the (unwanted) portion of the opposite channel is cancelled by that from the centre to restore the original sound stage. However, the L+R portion is still physically reproduced through the centre speaker, and is expected to improve dialogue / vocal clarity etc. The extent to which all this happens can also be controlled by the user.
Code:
m = (spl0 + spl1)*(1-slider1)*0.5;
s = (spl0 - spl1 )*(1+slider1)*0.5;
spl0 = m - s;
spl1 = m + s;The result of the matrix is:
Lout = L - slider*R
Cout = (L + R) * slider
Rout = R - slider*L
When played back through 3 identical speakers, the (unwanted) portion of the opposite channel is cancelled by that from the centre to restore the original sound stage. However, the L+R portion is still physically reproduced through the centre speaker, and is expected to improve dialogue / vocal clarity etc. The extent to which all this happens can also be controlled by the user.
Attachments
I tried the dual 3FE22 "Hydra" clones and the center image on some recordings was so great, that it made me think the LCR setup might be a bit overkill in this small room. It was almost like if the speakers were off and the sound was coming from the space between them.
Two things were obviously missing - bass and volume. I definitely need to build the second "Kallax" MEH or another symmetrical pair based on the planar driver soon.
Two things were obviously missing - bass and volume. I definitely need to build the second "Kallax" MEH or another symmetrical pair based on the planar driver soon.