At this point, I'm fairly convinced that Elias' idea isn't just a clever way to save space, it's like the BEST way to get excellent coherency and imaging.
In a nutshell, if you cram a stereo pair of speakers into a corner, the corner of the room forms a waveguide. (My loudspeakers use a waveguide, and I think this 'trick' is best with waveguide loaded drivers.)
Basically, the sound starts at the diaphragm of the driver, continues down the waveguide that's part of the loudspeaker, then continues down the waveguide formed by the room boundaries.
I elaborated on this idea here:
Audio Psychosis • View topic - A Soundstage with Width and Depth.
You can listen to this type of setup with ambio processing, but it sounds pretty darn good even without it. I've been listening without it.
I think people would be surprised by this setup; it's taught me that a lot of what we consider 'stereo' is simply an illusion introduced by the room. (For instance, creating additional width by cross firing speakers. It's an illusion, albeit a pleasant one.)
Hi all,
This SSSx5 is amazing. i am a believer of this layout. As 80ish% of music is in the center anyway.
I find the stereo triangle hard to listen to now. It seems so over separated and centre images dart every where. Thanks again Elias for the simple formula. Love it. You can get that live performance anywhere in the room.
Well done
This SSSx5 is amazing. i am a believer of this layout. As 80ish% of music is in the center anyway.
I find the stereo triangle hard to listen to now. It seems so over separated and centre images dart every where. Thanks again Elias for the simple formula. Love it. You can get that live performance anywhere in the room.
Well done
I have also been trying the SSS, almost OT because it's no more with the troika of 4" FR and not from a single speaker, but it's still SSS.
As it's a triamp, the SSS is only applied to the mids channels, but these mids allow many fantaisies till 12000 Hz if required, and at the opposite, the "tweeters" can go well under 1000 Hz, all this in a mouse click. The mid driver is a 12"PA in an array of 5 elements, actually not in Bessel but in power tapered with predominance of the central driver. Naturally as I have stock, the central driver will be the same, a couple of them paralleled, baffleless to go faster. All these details can have some importance, and are not definitive.
The SSS can be shunted at will with a good old cable. Speakers at 3 meters from the back wall, listener generally at 7 meters from the speakers.
first config:
I did very fast a side firing orthogonal. Average quality. With the SSS on, there is a very good first impression of sudden depth. Second impression,this only comes from the loss of level of the mids. Third impression, the mid channels don't give any lateral differentiation on music or pink noise. Left or right, it's the same. Even when asking the maximum from the tweeters it's never satisfying for stereo. Listened from close or far doesn't make any difference. Disaster.
second config:
Going back to the roots, this apparatus is more an OSD than a conventional triangle. For the same reasons, the depth is the first feeling, but this time, the stereo works, and there is something hard to describe, like a better extraction of the information, very evident on the voices. There is a lot of presence in the center, but there is also a greater dynamic of the stereo effect. For example when the guy comes closer to the mike (but it's not the unatural "wide mouth effect"). The background reverberations, as when speaking in a church are better perceived.
So, for me the SSS is in this case a real enhancement. Switching on/off makes no mercy for the normal system.
BTW what puzzles me is that the level of the mids channel is down ot a lot of dB with the SSS drivers in serie, but this is not perceived as a smile curve. The drivers are well in time, there is no cancellation. The central drivers are around -6dB relatively to the arrays, so they are barely audible. Of course without baffle, they start to roll of under 400 Hz, I did an EQ correction for this, but in no case rising up the level of the mid channel is positive. Strange.
Other detail, I have the feeling that the mid channel with SSS has to be low passed under 3000 Hz at least. Over this limit, there is something wrong, but this can be specific of this system.
Very positive idea Elias, thanks for the boost.
As it's a triamp, the SSS is only applied to the mids channels, but these mids allow many fantaisies till 12000 Hz if required, and at the opposite, the "tweeters" can go well under 1000 Hz, all this in a mouse click. The mid driver is a 12"PA in an array of 5 elements, actually not in Bessel but in power tapered with predominance of the central driver. Naturally as I have stock, the central driver will be the same, a couple of them paralleled, baffleless to go faster. All these details can have some importance, and are not definitive.
The SSS can be shunted at will with a good old cable. Speakers at 3 meters from the back wall, listener generally at 7 meters from the speakers.
first config:
I did very fast a side firing orthogonal. Average quality. With the SSS on, there is a very good first impression of sudden depth. Second impression,this only comes from the loss of level of the mids. Third impression, the mid channels don't give any lateral differentiation on music or pink noise. Left or right, it's the same. Even when asking the maximum from the tweeters it's never satisfying for stereo. Listened from close or far doesn't make any difference. Disaster.
second config:
Going back to the roots, this apparatus is more an OSD than a conventional triangle. For the same reasons, the depth is the first feeling, but this time, the stereo works, and there is something hard to describe, like a better extraction of the information, very evident on the voices. There is a lot of presence in the center, but there is also a greater dynamic of the stereo effect. For example when the guy comes closer to the mike (but it's not the unatural "wide mouth effect"). The background reverberations, as when speaking in a church are better perceived.
So, for me the SSS is in this case a real enhancement. Switching on/off makes no mercy for the normal system.
BTW what puzzles me is that the level of the mids channel is down ot a lot of dB with the SSS drivers in serie, but this is not perceived as a smile curve. The drivers are well in time, there is no cancellation. The central drivers are around -6dB relatively to the arrays, so they are barely audible. Of course without baffle, they start to roll of under 400 Hz, I did an EQ correction for this, but in no case rising up the level of the mid channel is positive. Strange.
Other detail, I have the feeling that the mid channel with SSS has to be low passed under 3000 Hz at least. Over this limit, there is something wrong, but this can be specific of this system.
Very positive idea Elias, thanks for the boost.
Attachments
Radugazon, you 2nd config is closer to the Gerzon 3 channel setup, including to a degree the hp/lp filters that Gerzon uses.
I was running the Gerzon 3 channel setup with outstanding success. I will be doing this again at some point. I just got to sort out the best implementation.
I dont really want to have a PC doing the dsp work. (as per my reaper code). You could do Gerzon's setup either passively (2 x M/S transfomers + hp /lp trickery) or active with opamps. Both are beyond my skills at this point. I actually know how to implement. Just cant do it...
Great experiment tho. Thanks for posting...
Optic
I was running the Gerzon 3 channel setup with outstanding success. I will be doing this again at some point. I just got to sort out the best implementation.
I dont really want to have a PC doing the dsp work. (as per my reaper code). You could do Gerzon's setup either passively (2 x M/S transfomers + hp /lp trickery) or active with opamps. Both are beyond my skills at this point. I actually know how to implement. Just cant do it...
Great experiment tho. Thanks for posting...
Optic
Wow !!
I agree with optic that the 2nd setup is not a 'single' speaker any more. (Technically, the 1st one isn't, either. Arrays and modular everything, whew~ Oh well.... )
Anyway, the 2nd setup is more of linear matrix 3-ch, but not only that. You got a complex recipe here.
Already very impressive now, but I'm also wondering what'll happen if your IB and OB woofers are also wired as linear matrix 3-ch. Don't know the effect in LF.
I agree with optic that the 2nd setup is not a 'single' speaker any more. (Technically, the 1st one isn't, either. Arrays and modular everything, whew~ Oh well.... )
Anyway, the 2nd setup is more of linear matrix 3-ch, but not only that. You got a complex recipe here.
Already very impressive now, but I'm also wondering what'll happen if your IB and OB woofers are also wired as linear matrix 3-ch. Don't know the effect in LF.
Can you develop more on this Gerzon 3 channels please? I have been searching in the archive, but I'm confused.
I already tried many setups with rear channels, but the SSS if far simpler to implement at first sight. The complications come when it's time to compare subjectively different slopes and different Fc. Any modification affects width and localizations or induce some spatial distorsions. Here, the use of a DSP (DCX in my case) can be seen as a luck or a curse.
An other interrogation is for me the equalisation of the system. Once the matrix is on, it's impossible to isolate a channel for measuring it normally. A thing is sure anyway, the SSS doesn't harm the time cohesion of the system. This step comes from a kind of LR 12 @ 3000, it could have been much worse.
About the pertinence of the extension to the subs, I dunno because the arrays go to 100 Hz already but the central speaker never goes so low, and it already gives good results. Maybe there is a privilegiated band ( exactly the telephone band...) where this works the best. I dunno but I could give a try to this idea.
I already tried many setups with rear channels, but the SSS if far simpler to implement at first sight. The complications come when it's time to compare subjectively different slopes and different Fc. Any modification affects width and localizations or induce some spatial distorsions. Here, the use of a DSP (DCX in my case) can be seen as a luck or a curse.
An other interrogation is for me the equalisation of the system. Once the matrix is on, it's impossible to isolate a channel for measuring it normally. A thing is sure anyway, the SSS doesn't harm the time cohesion of the system. This step comes from a kind of LR 12 @ 3000, it could have been much worse.
About the pertinence of the extension to the subs, I dunno because the arrays go to 100 Hz already but the central speaker never goes so low, and it already gives good results. Maybe there is a privilegiated band ( exactly the telephone band...) where this works the best. I dunno but I could give a try to this idea.
Attachments
Last edited:
I found this:
Upward Mobility: From 2 Channels to Surround | Stereophile.com
There're some more names and references following that. I haven't read them all, though...
And this from an old thread:
http://www.diyaudio.com/forums/mult...t-channels-out-stereo-signal.html#post2660446
However, Markus himself seems not interested in this technique. I myself like it very much.
Upward Mobility: From 2 Channels to Surround | Stereophile.com
There're some more names and references following that. I haven't read them all, though...
And this from an old thread:
http://www.diyaudio.com/forums/mult...t-channels-out-stereo-signal.html#post2660446
However, Markus himself seems not interested in this technique. I myself like it very much.
Last edited:
Hi CLS, following your suggestion I have been trying the SSS on the subs. Paralleled OB going through the IB, this last probably in serie parallel.
I was not expecting this but something happens. In despite of their low pass IV th order under 100 Hz, when listening the subs channel alone, they can reproduce very progressively the translation of an appropriate source as a train passing. When listening the whole system, on a few records some events previously very externalized are now more centered, and the opposite happens with some others.
Better or worse, I dunno what to say, but it's clear that the SSS works also under 100 Hz.
I was not expecting this but something happens. In despite of their low pass IV th order under 100 Hz, when listening the subs channel alone, they can reproduce very progressively the translation of an appropriate source as a train passing. When listening the whole system, on a few records some events previously very externalized are now more centered, and the opposite happens with some others.
Better or worse, I dunno what to say, but it's clear that the SSS works also under 100 Hz.
Hi Radugazon,
Give us a day or so. I will write up my methods and its implementation. i.e gerzon 3 channel. I have read the patients a million times now. I will do my best to describe them and how i did this.
Cheers,
Optic...
So true. Best loudspeaker system utilises the room as a benefit and not try to fight against it.
About the corner placement.. while it integrates low and mid freqs very well into the room, there is one problem for me that it does not solve: high freqs. Depending on the speaker high freq directivity, and especially regarding corner placement a horn type of speaker is commonly used meaning relatively high directivity, the high freqs at the listening position causes the normal interference field combing pattern of the conventional stereo triangle which in turn causes the speakers to become localisable themshelves not allowing realistic high freq phantom images to form.
This above problem of stereo triangle was the main motivator to find a solution. And there is a solution !
Simply, do not let an interference field to be formed at the listening position in such a manner that combing is perceivable not even with lateral head movements and head turning. This also ensures no sweet spot exists.- Elias
Me too. I still have a normal stereo speakers in my room, and for some masochist reason I like to switch them on occasionally just to find out all the problems of stereo triangle
The freedom of no sweet spot of the SSSx5 has spoiled me !
I'm not anymore willing to sit in a tiny spot and with head kept forward all the time while listening ! I find it absurd, actually, that people let themselves to be submitted to the authority of a stereo triangle, like they would have no will of they own 
Even my dipole line arays, do I dare to say this, cannot compete with SSSx5 in the level of realistic imaging and spaciousness
Dipole line arrays do excel extremely well at low and low-mid freqs in a small room, but in overall they lack the capability of forming a believable reproduction of the real event. This only appears to indicate that in a perception of sound in a small room acoustic space, the high-mid and high freqs completely dominate everything: both the imaging and spaciousness.- Elias
I agree that spaciousness is a nice effect but that's all it is, a effect. I use my system for movies, TV and music so spaciousness is just one desirable feature amongst others. The single speaker system discussed here creates a nice effect with a few recordings but it's not a practical nor a desirable solution for a multi purpose system.
Interesting "single speaker" developments: Cambridge Mechatronics
Interesting "single speaker" developments: Cambridge Mechatronics
Everything you do is so uncomparable
Just a thought.. In a system with very high level of complexity, one can lose focus. I don't doubt your capabilities, not at all, but because everything about psychoacoustics is not known. Too many simultaneously varying factors means confusion of what is important what is not and which is the right direction. Only a fearless experimenter like you can tackle the problem.
There seems to be something common with my findings too, I use the psychoacoustic filter somewhere above 3kHz to attenuate center element.
Also this idea seems not to be a random lottery since Gerzon is using similarly motivated filter in Trifield.
From papers of Ambisonics and Trifield (I should have most of them
) one can say that perception is based on two (or more) regions: Below 700Hz is so called 'velocity' region and between 700Hz to 5kHz is 'energy' region. Above 5kHz is something else, I presume pinna localisation becomes dominate.Below 700Hz the velocity vector defines the localisation perception. Basically the sound is based on vector summation which is possible because of long wavelength.
Above 700Hz phase becomes ambigous and intensity must be used defining the energy vector determining the localisation.
When getting into the equations it gets complex ! There are so many configuration of speaker layouts, optimisation principles, assumed recording techniques and evolutions of the same principle etc that I would consider it close to impossible to definitely say what it the 'best'. However, for an experimenter it's a fruitful playground
A starting point maybe:
Sound reproduction system having a matrix converter
Well, it's a patent, and as we know patents contain 'errors' to prevent direct copying
Let's see what Optic will come up with
I'm too interested.- Elias
Is this observation based solely on the miniature cardboard box of yours from post #252 ?
In another board there was a complete forum dedicated on these kind of bar speakers. I was surprised, it seems to be a trend going on of integrating speakers in a singular point. Look at the all ipod gadgets, stereo sound from a singular point.
Who knows where is this leading to..
- Elias
My observation is based on that one and multiple incarnations utilizing the same principle. It really has too many drawbacks. The solution is more speakers.
How about SSS + Watson ?
-------
As to the sound bar, it seems Yamaha has done it and still doing it with great success. It's said the system relies on DSP. I believe so, especially those commercial products are equipped with so many functions (operating modes).
It doesn't appear that I've chimed in on this thread, although I could have sworn I had.....
I tried something pretty close to SSS from what I can tell. I took 3 JBL Control Now speakers and arranged them in to a 270-degree arc on top of the TV. Being limited to what the receiver supported, I ran DPLIIx to create the synthesized center and re-process the L/R. Generally speaking, it worked pretty well although I definitely can't say it was any kind of a revelation for high-quality music listening. Served it's purpose though, which was to give acceptable sound from a 'single speaker' atop the TV.
The Control Now speakers are well suited for this type of thing due to their shape, although the radiation pattern and degree of baffling between the L/R don't quite match Elias' setup.
For lack of a 'psychoacoustics' forum, I'll throw an additional more or less OT but vaguely related thought or 2 in.....
I've recently become rather intrigued by the Smyth Realiser (www.smyth-research.com). It's a little DSP device that uses in-ear binaural mics to measure the in-room binaural impulse response of a real speaker setup using your own HRTF, and then runs a convolution engine over them to provide playback over headphones for stereo or 5.1/7.1 sources. The fundamental idea isn't necessarily revolutionary (I had the same idea for stereo 12+ years ago when business travel kept me away from my first successful DSP/FIR speaker setup) but what is remarkable is the near-universal opinion of reviewers/owners that it really actually delivers on the idea - it really sounds like you're listening to speakers in a room. Many folks have gone to AIX studios (they master blu-ray audio for Hollywood) and gotten measured in their $250k+ mastering studio, and now can watch movies as though they're at the control desk.
My initial reaction to it was to be a bit surprised that they don't do more to eliminate/correct for the problems of the speakers and room - that in fact it seems to be the entire point to include those in the response. Part of this is undoubtedly marketing, but I wonder whether a large part is also that a 'real' reverberant field is a critical part of creating a believable sense of space. This would imply that synthetic reverb added back in to a sterilized/corrected version of the binaural impulses doesn't cut it.
My second reaction is that this would seem to be an absolutely ideal way to implement ambiophonics - 'simply' measure a single high-quality mono speaker directly in front, and then separate the channels to eliminate the x-talk component - you get a real measured HRTF response and perfect x-talk cancellation. I'm so intrigued by this idea that I ordered an in-ear binaural mic setup to play with.
I tried something pretty close to SSS from what I can tell. I took 3 JBL Control Now speakers and arranged them in to a 270-degree arc on top of the TV. Being limited to what the receiver supported, I ran DPLIIx to create the synthesized center and re-process the L/R. Generally speaking, it worked pretty well although I definitely can't say it was any kind of a revelation for high-quality music listening. Served it's purpose though, which was to give acceptable sound from a 'single speaker' atop the TV.
The Control Now speakers are well suited for this type of thing due to their shape, although the radiation pattern and degree of baffling between the L/R don't quite match Elias' setup.
For lack of a 'psychoacoustics' forum, I'll throw an additional more or less OT but vaguely related thought or 2 in.....
I've recently become rather intrigued by the Smyth Realiser (www.smyth-research.com). It's a little DSP device that uses in-ear binaural mics to measure the in-room binaural impulse response of a real speaker setup using your own HRTF, and then runs a convolution engine over them to provide playback over headphones for stereo or 5.1/7.1 sources. The fundamental idea isn't necessarily revolutionary (I had the same idea for stereo 12+ years ago when business travel kept me away from my first successful DSP/FIR speaker setup) but what is remarkable is the near-universal opinion of reviewers/owners that it really actually delivers on the idea - it really sounds like you're listening to speakers in a room. Many folks have gone to AIX studios (they master blu-ray audio for Hollywood) and gotten measured in their $250k+ mastering studio, and now can watch movies as though they're at the control desk.
My initial reaction to it was to be a bit surprised that they don't do more to eliminate/correct for the problems of the speakers and room - that in fact it seems to be the entire point to include those in the response. Part of this is undoubtedly marketing, but I wonder whether a large part is also that a 'real' reverberant field is a critical part of creating a believable sense of space. This would imply that synthetic reverb added back in to a sterilized/corrected version of the binaural impulses doesn't cut it.
My second reaction is that this would seem to be an absolutely ideal way to implement ambiophonics - 'simply' measure a single high-quality mono speaker directly in front, and then separate the channels to eliminate the x-talk component - you get a real measured HRTF response and perfect x-talk cancellation. I'm so intrigued by this idea that I ordered an in-ear binaural mic setup to play with.
Ambiophonics sacrifices the L and R HRTF for getting the C HRTF right. So capturing a center HRTF for a headphone based playback system is just the first step. What you could and should do is capture the HRTFs for the locations of L, C and R. Then derive a L, C and R channel (with phantom images between L/C and R/C only), feed through the appropriate HRTFs and sum to a L and R headphone signal.
A more spacious presentation could be obtained by deriving wide and height information for the front and the back and feeding it through appropriate HRTFs.