Hi there,
Let's have a revisit to this. In post #299, there's a 4-ch proposal:
I seems simple, just an ordinary stereo amp and 4 speakers without any fancy filter. Has anyone tried it?
I've tried with a pair of Foxtex's small monitors and a pair of Jordan's fullranger (pointed by red arrows):
With other things in the photo staying quiet, I've been listening to these 4 little guys for more than 2 days. I like it.
I can't get the outer pair to 55 degree each side, only 37 degree at the max. And it's 15 degree each on the inner pair. Nevertheless, it projects a huge space and very good envelopment.
The fostex is active. I use another amp with independent level control for Jordan. So I can adjust the balance between inner main channels and differential side channels. It can be overdone with too much differential and sounding hollow and unnatural.
In the process of setting, once I detected comb-filtering when I moved straight fore-aft. I have never had such experience. I think it's coming from the overlapped portion between inner and outer pairs. But interestingly I can't bring it back and examine again.
An important factor for me is the central imaging. For this, new 4-ch setup performs as good as 3-ch linear matrix previously used. It's also natural when listening off center.
.........
Up to this point, I'm thinking the tiny delay in the RACE. The cross-feeding inverse bandpassed (250-5kHz) signal is delayed by up to 100 microsec. It's the distance of 3.44 cm (for soundwave, of course). I wonder, can we really detect that?
Let's have a revisit to this. In post #299, there's a 4-ch proposal:
An externally hosted image should be here but it was not working when we last tested it.
I seems simple, just an ordinary stereo amp and 4 speakers without any fancy filter. Has anyone tried it?
I've tried with a pair of Foxtex's small monitors and a pair of Jordan's fullranger (pointed by red arrows):
With other things in the photo staying quiet, I've been listening to these 4 little guys for more than 2 days. I like it.
I can't get the outer pair to 55 degree each side, only 37 degree at the max. And it's 15 degree each on the inner pair. Nevertheless, it projects a huge space and very good envelopment.
The fostex is active. I use another amp with independent level control for Jordan. So I can adjust the balance between inner main channels and differential side channels. It can be overdone with too much differential and sounding hollow and unnatural.
In the process of setting, once I detected comb-filtering when I moved straight fore-aft. I have never had such experience. I think it's coming from the overlapped portion between inner and outer pairs. But interestingly I can't bring it back and examine again.
An important factor for me is the central imaging. For this, new 4-ch setup performs as good as 3-ch linear matrix previously used. It's also natural when listening off center.
.........
Up to this point, I'm thinking the tiny delay in the RACE. The cross-feeding inverse bandpassed (250-5kHz) signal is delayed by up to 100 microsec. It's the distance of 3.44 cm (for soundwave, of course). I wonder, can we really detect that?
Last edited:
This is the Hafler circuit, or Dynaquad, from the 70s.
Hafler circuit - Wikipedia, the free encyclopedia
Ambiophonics/Envelophonics
If you want a full circle of 3D sound from 4.0 (5.1 media or an up to 180 degree wide stage from 2.0 media like existing CDs and LPs then you just need four speakers (nothing to do with Hafler or Quad) arranged as one Ambiodipole in front of you and one directly behind. An Ambiodipole is two speakers about 20 degrees apart fed from a 2.0 source that is processed by a program like RACE, BACCH or similar so that the crosstalk between the speakers is eliminated recursively. (loudspeaker binaural) The rear Ambiodipole is fed the same file but through a second RACE processor so that the settings can be optimized for a different angle or speaker type and seasoned to taste. The rear pair can be closer behind than the front pair and its level about 6 dB less but nothing is critical. Just adjust the rear pair level and RACE settings for the greatest sense of space, stage width and depth. This rear Ambiodipole can then be alternatively used to produce a rear stage for movies that have one or music in the round.
The use of the rear Ambiodipole with 2.0 sources is called Envelophonics because it provides a feeling of concert hall or spatial realism called that by several AES researchers over the years. You can read about RACE and Envelophonics on the Ambiophonics.org website. It is an .org because like stereophonics, Ambiophonics is in the pubic domain and its architecture can be used by anyone interested in making Ambio components or programs.
If you want a full circle of 3D sound from 4.0 (5.1 media or an up to 180 degree wide stage from 2.0 media like existing CDs and LPs then you just need four speakers (nothing to do with Hafler or Quad) arranged as one Ambiodipole in front of you and one directly behind. An Ambiodipole is two speakers about 20 degrees apart fed from a 2.0 source that is processed by a program like RACE, BACCH or similar so that the crosstalk between the speakers is eliminated recursively. (loudspeaker binaural) The rear Ambiodipole is fed the same file but through a second RACE processor so that the settings can be optimized for a different angle or speaker type and seasoned to taste. The rear pair can be closer behind than the front pair and its level about 6 dB less but nothing is critical. Just adjust the rear pair level and RACE settings for the greatest sense of space, stage width and depth. This rear Ambiodipole can then be alternatively used to produce a rear stage for movies that have one or music in the round.
The use of the rear Ambiodipole with 2.0 sources is called Envelophonics because it provides a feeling of concert hall or spatial realism called that by several AES researchers over the years. You can read about RACE and Envelophonics on the Ambiophonics.org website. It is an .org because like stereophonics, Ambiophonics is in the pubic domain and its architecture can be used by anyone interested in making Ambio components or programs.
Thanks a lot for the response.
Pity that it's not practical for me to use rear speakers. And I can't bring the speakers near enough to utilise real Ambiophonics.
As my understanding of RACE, the principle is cancelling some signal of the opposite side -- a bandpassed (250-5kHz) signal is inverse and delayed a little bit, then mixed with the signal of the other channel. Vice versa.
I believe the 4-ch arrangement above is similar in cancelling some portion of the other side. There're two differences by this rough version:
1. No (proper) delay
2. The cancelling signal is from a different source
About delay, it can be done physically by the position of the separate speakers. But there's a draw back that there'd be smear between the speaker in the portion which should be the same on inner and outer speaker.
OTOH, I wonder how significant the delay could be. After all, micro-sec is a very small time and distance. How the effectiveness maintains when the listener is not perfectly at the center?
Pity that it's not practical for me to use rear speakers. And I can't bring the speakers near enough to utilise real Ambiophonics.
As my understanding of RACE, the principle is cancelling some signal of the opposite side -- a bandpassed (250-5kHz) signal is inverse and delayed a little bit, then mixed with the signal of the other channel. Vice versa.
I believe the 4-ch arrangement above is similar in cancelling some portion of the other side. There're two differences by this rough version:
1. No (proper) delay
2. The cancelling signal is from a different source
About delay, it can be done physically by the position of the separate speakers. But there's a draw back that there'd be smear between the speaker in the portion which should be the same on inner and outer speaker.
OTOH, I wonder how significant the delay could be. After all, micro-sec is a very small time and distance. How the effectiveness maintains when the listener is not perfectly at the center?
The angle between the speakers is not all that critical. Just bring them in as much as you can. It will still be better than normal stereo since a lot of the crosstalk will still be cancelled.
The signal does not need to be band pass filtered. RACE now works full range. My earlier papers suggesting this were just to be sure that if there were problems at the frequency extremes for audiophiles that they would be able to bypass them. There is no crosstalk at the low bass or above say 7000Hz so RACE need not operate there. But there is no harm if it does and now most of the apps are full range.
It is not a simple delay. It is a recursive feedback delay process that continues until each crosstalk cancelling sample declines to inaudibility. The many delays and the decreases in the amplitude of each sample is set by the listener by ear based on the angle to the speakers and is not critical. The cancelling signals are from the same speakers as the direct sound. If the crosstalk is eliminated than by definition you end up with a perfect left channel signal at the left ear and a perfect right channel signal at the right ear, just as with earphones but from loudspeakers, thus loudspeaker binaural.
You cannot do this just by using a separate speaker as Polk did years ago. This method is not recursive and cannot be adjusted. It is like the primitive Sonic Holography and Lexicon one shot cancellers used before DSP made a better real time process possible.
With a loudspeaker binaural system a change of 10 microseconds by moving a speaker an inch or so is audible as a change in stage width or symmetry. But with the speakers somewhat in front, you can move back and forth, rotate, nod or lean the head, stand or lie down and actually will have less of a problem than with stereo. In stereo if you move forward you get a hole in the middle. If you move back you get mono. If you go to the side you get one channel, mostly. In Ambio if you move to the side you still get both channels clearly audible, so in 5.1 you never need a center speaker. If you move up toward the speakers you just get stereo. If you move back almost nothing happens.
Try Ambiophonics with two or four speakers. It costs next to nothing.
The signal does not need to be band pass filtered. RACE now works full range. My earlier papers suggesting this were just to be sure that if there were problems at the frequency extremes for audiophiles that they would be able to bypass them. There is no crosstalk at the low bass or above say 7000Hz so RACE need not operate there. But there is no harm if it does and now most of the apps are full range.
It is not a simple delay. It is a recursive feedback delay process that continues until each crosstalk cancelling sample declines to inaudibility. The many delays and the decreases in the amplitude of each sample is set by the listener by ear based on the angle to the speakers and is not critical. The cancelling signals are from the same speakers as the direct sound. If the crosstalk is eliminated than by definition you end up with a perfect left channel signal at the left ear and a perfect right channel signal at the right ear, just as with earphones but from loudspeakers, thus loudspeaker binaural.
You cannot do this just by using a separate speaker as Polk did years ago. This method is not recursive and cannot be adjusted. It is like the primitive Sonic Holography and Lexicon one shot cancellers used before DSP made a better real time process possible.
With a loudspeaker binaural system a change of 10 microseconds by moving a speaker an inch or so is audible as a change in stage width or symmetry. But with the speakers somewhat in front, you can move back and forth, rotate, nod or lean the head, stand or lie down and actually will have less of a problem than with stereo. In stereo if you move forward you get a hole in the middle. If you move back you get mono. If you go to the side you get one channel, mostly. In Ambio if you move to the side you still get both channels clearly audible, so in 5.1 you never need a center speaker. If you move up toward the speakers you just get stereo. If you move back almost nothing happens.
Try Ambiophonics with two or four speakers. It costs next to nothing.
From your picture, the small speakers are already close enough to be within the recommended less than 30 degrees.
My understanding of RACE is - there are many inverted signals each slightly attenuated than the other cancelling each other.
Btw, the rear speakers can be anywhere as long they are within the 30 degrees spread. Mine are located close to the ceiling facing my sitting position about 6/10 th distance of the front speakers distance.
Thanks for your patience in explaining all these again and again.
I guess I can't make the 'recursive feedback delay' myself. But even then, the crude 4-ch setup above do sound good. It needs some minor adjustments to different types of recordings. Nevertheless it's better than plain stereo.
It's sad that the D in D out miniambio is not coming. And I'm not ready to add a computer into my (rather complex) system, yet.
I guess I can't make the 'recursive feedback delay' myself. But even then, the crude 4-ch setup above do sound good. It needs some minor adjustments to different types of recordings. Nevertheless it's better than plain stereo.
It's sad that the D in D out miniambio is not coming. And I'm not ready to add a computer into my (rather complex) system, yet.
Oh well, I hope this is not OT too much. I'm not stubbornly refusing "real" ambiophonics, but before I can properly connect a computer to my audio system, I'd like to explore more on the Hafler circuit.
Last weekend, I re-arranged my speakers units to this:
(it's not easy to move them around, took me several hours of hard work)
Main channels are now baffle-less arrays and outer dipole subs, side channels are still the Jordan Watt Flagon. (DML panel is removed.)
OB arrays give superior performance than the small monitors in every aspect. In addition, I feel there're two other factors making the sound more intense and dynamic compared to the previous 3-ch setup:
1. closer distance between the two arrays (and the character is different from DML)
2. no reduction (process) in signal
I'm still kind of restless in tuning here & there, but I like the overall presentation. And I believe there's potential. It's well said by Pano in the previous post:
In ordinary stereo, phantoms are around the center; in ambiophonics, phantoms are outside.
In this 4-ch setup, there're side channels so it's not really phantom outside. But in proper setting (and proper recordings), all 4 speakers can play disappearing act. I also like it very much that the closer main channels make the central imaging very solid and more life-like.
This will be evolving. My plan is changing the bass to dipole Karlson (KaZBa), and DML panels for side channels. Finally, maybe bring back my Oris horn for main channel...
Last weekend, I re-arranged my speakers units to this:
(it's not easy to move them around, took me several hours of hard work)
Main channels are now baffle-less arrays and outer dipole subs, side channels are still the Jordan Watt Flagon. (DML panel is removed.)
OB arrays give superior performance than the small monitors in every aspect. In addition, I feel there're two other factors making the sound more intense and dynamic compared to the previous 3-ch setup:
1. closer distance between the two arrays (and the character is different from DML)
2. no reduction (process) in signal
I'm still kind of restless in tuning here & there, but I like the overall presentation. And I believe there's potential. It's well said by Pano in the previous post:
In ordinary stereo, phantoms are around the center; in ambiophonics, phantoms are outside.
In this 4-ch setup, there're side channels so it's not really phantom outside. But in proper setting (and proper recordings), all 4 speakers can play disappearing act. I also like it very much that the closer main channels make the central imaging very solid and more life-like.
This will be evolving. My plan is changing the bass to dipole Karlson (KaZBa), and DML panels for side channels. Finally, maybe bring back my Oris horn for main channel...
OK, I tried it, just now, finally.
I temporarily connect my old lap top to Creative sound blaster digital music sx by USB, and its S/PDIF output to my DEQ2496 in the system.
I use Hotto's ambiophonics player to play .wav files in the PC.
First I try pure left & right signals provided by that site. Yes the sounds extend beyond the positions of speakers, but what I hear is a smeared area crossing the angle from 15 degree to 40 degree a side. I'm not sure if this is correct (optimal) or not.
USB cable is not long enough, so I can only put the lap top near equipment racks. For operating the PC and listening, I move fore-aft a lot. When doing so, I hear an effect similar to comb filtering. This is stronger than what I experienced previously when doing 4-ch Hafler circuit trial in the first time.
I've listened to some familiar tracks of different genres (originally from different CDs, of course). In some of them, the effect is awesome right away. Very wide and deep sound stage with precise imaging.
But in some other recordings, I encounter the following problems:
1. Weak and lean central imaging of the singer in pop music. And it seems the singer is standing too far behind other sounds. I try increasing 'Center%', but the improvement is limited.
2. Strange feel of envelopment. Sounds seem bending towards me from two sides, but it's not the relocation of sound images. It's like there's some kind of halo or air spreading from the sides and surround me.
3. Together with the sense of strong envelopment, there's always a significant change in tonal balance -- a broad bump in the lower mid to upper bass. It also kind of feels like putting hands around ears -- but now the hands are huge and about 1m away from my ears.
4. Sometimes the apparent size of bass sound would expand too much and almost cover a half of wall. And the related level is bumped significantly, too.
In the process, I can't go longer delay than the default 91 micro-sec., or there would be strange feels. Less is usually more natural to me. And fullrange is not good in my system, there would be severe hiss noise. Making it under 4.5kHz avoids that problem.
Not sure by now, I think possible reasons would be:
1. Too much reflections to mess up with the proper cross talk cancellation, as there's barely any acoustic treatment.
2. Improper setting of the amount of XTC and delay time. I need more time for this.
3. The bass units are too far from the main speakers.
Nevertheless, this brief first trial reveals a very high potential. I'd need a proper hardware for this. A noisy PC with inconvenient interface is not a long term solution as it needs a lot of fine tunings.
Of course I make some comparisons with the 4-ch Hafler circuit. It's not too shabby and also fairly spacious sounding, but not as astonishing in the sheer size of ambiophonic's sound field. In the situations I can't properly setup the computer, the old Hafler stays natural sounding.
Can't help thinking, it'd not be ideal in every recording. And when the unnatural effects are dialed out by optimal tuning, the magic of ambiophonic might also go away...
I need more time to find out.
Mr. Ralph Glasgal, thank you very much.
I temporarily connect my old lap top to Creative sound blaster digital music sx by USB, and its S/PDIF output to my DEQ2496 in the system.
I use Hotto's ambiophonics player to play .wav files in the PC.
First I try pure left & right signals provided by that site. Yes the sounds extend beyond the positions of speakers, but what I hear is a smeared area crossing the angle from 15 degree to 40 degree a side. I'm not sure if this is correct (optimal) or not.
USB cable is not long enough, so I can only put the lap top near equipment racks. For operating the PC and listening, I move fore-aft a lot. When doing so, I hear an effect similar to comb filtering. This is stronger than what I experienced previously when doing 4-ch Hafler circuit trial in the first time.
I've listened to some familiar tracks of different genres (originally from different CDs, of course). In some of them, the effect is awesome right away. Very wide and deep sound stage with precise imaging.
But in some other recordings, I encounter the following problems:
1. Weak and lean central imaging of the singer in pop music. And it seems the singer is standing too far behind other sounds. I try increasing 'Center%', but the improvement is limited.
2. Strange feel of envelopment. Sounds seem bending towards me from two sides, but it's not the relocation of sound images. It's like there's some kind of halo or air spreading from the sides and surround me.
3. Together with the sense of strong envelopment, there's always a significant change in tonal balance -- a broad bump in the lower mid to upper bass. It also kind of feels like putting hands around ears -- but now the hands are huge and about 1m away from my ears.
4. Sometimes the apparent size of bass sound would expand too much and almost cover a half of wall. And the related level is bumped significantly, too.
In the process, I can't go longer delay than the default 91 micro-sec., or there would be strange feels. Less is usually more natural to me. And fullrange is not good in my system, there would be severe hiss noise. Making it under 4.5kHz avoids that problem.
Not sure by now, I think possible reasons would be:
1. Too much reflections to mess up with the proper cross talk cancellation, as there's barely any acoustic treatment.
2. Improper setting of the amount of XTC and delay time. I need more time for this.
3. The bass units are too far from the main speakers.
Nevertheless, this brief first trial reveals a very high potential. I'd need a proper hardware for this. A noisy PC with inconvenient interface is not a long term solution as it needs a lot of fine tunings.
Of course I make some comparisons with the 4-ch Hafler circuit. It's not too shabby and also fairly spacious sounding, but not as astonishing in the sheer size of ambiophonic's sound field. In the situations I can't properly setup the computer, the old Hafler stays natural sounding.
Can't help thinking, it'd not be ideal in every recording. And when the unnatural effects are dialed out by optimal tuning, the magic of ambiophonic might also go away...
I need more time to find out.
Mr. Ralph Glasgal, thank you very much.
Last edited:
I've tried that tool as others and I've lived with ambiophonics for 7 years already with a wide variety of speakers and rooms and barriers.
Trust me when I tell you, if you want Ambiophonics, use a thick physical barrier with an absorptive surface. The digital crosstalk cancellators are terrible and they don't even work properly, they are a joke. You could use them on cheap speakers and at close range <1m if you really want to...
If you want good sound, the physical barrier is the way to go. If you're not willing to divide your room with a massive physical barrier for the sake of sound: FORGET IT!!
Trust me when I tell you, if you want Ambiophonics, use a thick physical barrier with an absorptive surface. The digital crosstalk cancellators are terrible and they don't even work properly, they are a joke. You could use them on cheap speakers and at close range <1m if you really want to...
If you want good sound, the physical barrier is the way to go. If you're not willing to divide your room with a massive physical barrier for the sake of sound: FORGET IT!!
Yes, and even a smallish panel in front of your head (but fairly close to your nose) does a lot.
Make sure you use absorption on the panel if you want most natural sound. I've tried Metal, Cardboard, Wood u name it. Reflective materials widen the stage slightly cos u have more contrast between ambient sound and direct sound per ear but absorption creates less distortions.
The V or U shaped sound stage however is a natural problem of ambio (Meaning that the further the sounds locate off-center the closer they appear) But that's no fault of ambiophonics, that's cos ur playing music not suitable for an ambio setup.
The ILD increases towards higher frequencies naturally cos higher frequencies coming in from the side lose more energy moving around your head to the other ear than lower notes. Since no recording studio in the world cares about that little fact, the only conclusion your brain takes from an unnaturally high ILD is that the source is very close to your position. The brain doesn't locate sounds below 1khz using ILD's. ILD's below 1khz rather determine the distance of a source starting at no further than 2 metres.Also clean sounds with little harmonics tend to move around depending on pitch.
If bad ILD's are mixed with bad ITD's and inept reverb which usually is the case you get S***. Welcome to the real world.
Back up a foot or two from the barrier; soundstage narrows - problem solved. Not ideal but still better than Stereo. Good luck trying to fiddle with the knobs in frustration if you're using some Crosstalk Cancellation Software.
The V or U shaped sound stage however is a natural problem of ambio (Meaning that the further the sounds locate off-center the closer they appear) But that's no fault of ambiophonics, that's cos ur playing music not suitable for an ambio setup.
The ILD increases towards higher frequencies naturally cos higher frequencies coming in from the side lose more energy moving around your head to the other ear than lower notes. Since no recording studio in the world cares about that little fact, the only conclusion your brain takes from an unnaturally high ILD is that the source is very close to your position. The brain doesn't locate sounds below 1khz using ILD's. ILD's below 1khz rather determine the distance of a source starting at no further than 2 metres.Also clean sounds with little harmonics tend to move around depending on pitch.
If bad ILD's are mixed with bad ITD's and inept reverb which usually is the case you get S***. Welcome to the real world.
Back up a foot or two from the barrier; soundstage narrows - problem solved. Not ideal but still better than Stereo. Good luck trying to fiddle with the knobs in frustration if you're using some Crosstalk Cancellation Software.
I have used panels and AmbiophonicsDSP and I think the DSP version is more flexible. You need to get the speakers separation with 25 degrees. All the V and U shape soundstage are usually the product of studio recordings. The best way to implement Ambiophonics, IMHO, is Electro-music's AmbiophonicsDSP. Recordings such as Tracy Chapman can be realistically reproduced with Ambiophonics using the DSP with the space function almost the max. You can also bring the centre in or out using the centrum function.
Please note that the delay value changes automatically according to the file sampling rate. I am now using JRiver to output all files at 32/88.2 so that the delay remains the same irrespective of the file format.
Please note that the delay value changes automatically according to the file sampling rate. I am now using JRiver to output all files at 32/88.2 so that the delay remains the same irrespective of the file format.
I finally built some speakers with the HiVi Trinity 6 triaxials and, aimed away from the listener's position, they image much better than the ones I have been using so far. As always, I use the Ambio4you, but the physical barrrier is now very small yet the soundstage is wider than before.
Attachments
This is interesting. I too did try with physical barrier and Miniambio but did not really perceive any difference in imaging. There was some difference in the overall sound. Currently, just using Filmakers AmbiophonicsDSP. Could you share audio sample of the actual sound in the room?
These are videos of my setups sound.
https://www.youtube.com/watch?v=7RYoxKYKNV4
https://www.youtube.com/playlist?list=PLdMpXpnMH1M8rI-qlCRLZsPALIe8Ja8zA
Are these recorded binaurally?
Here is a binaural recording if my set up 4 years ago.
https://m.youtube.com/watch?v=7cVOI4-bj6g
The speaker is called the "event horizon"
There is a build thread here.
Btw, sweet soundlabs!