[opaqueice]

Please note, that we are speaking about 2 things, the reproduction of the first arrival of the direct sound and recreating an ambient filed similar to the original.
The direct sound is reproduced through XTC and the ambient filed  - if it is not recorded independently - created by simulation. With the proper mix of the two, you can get surprisingly "real" reproduction of real acoustic events.
Of course with studio recordings ( drums in an anechoic booth, guitars through pickup, vocal close mic-ed and with different added echo in every line ) it is a hit and miss.

OK, but I'm trying to understand why you think the direct sound comes from such a narrow angle?  In many live venues it doesn't, including concert halls when you're seated near the front.

[sunshinedawg]

OK, but I'm trying to understand why you think the direct sound comes from such a narrow angle?  In many live venues it doesn't, including concert halls when you're seated near the front.

I think the short answer is you just have to try what we're doing. You don't have to worry about where you are seated. The techniques we are using address all of this. In fact, through tweaking the delay and the strength of the ambient channels, you can make it sound like you are anywhere in a hall, from the first row, to the balcony.

It is interesting to note that, by having the mains 10o apart, I can get sounds spanning about 160o in front of me, with my ambient channels turned off. The best an equilateral setup can do is put the image between the speakers. You might claim that the sounds come from outside the speakers sometimes, but this is just an anomaly, it is not reproducible to any degree.

This what an equilateral setup can do: triangle restriction

Here's what Frank and my sytem can do: crosstalk cancelled

We get a very wide image, way wider than with a triangle.

[Daygloworange]

Just follow the instructions in the first one for speaker setup: " Ideally, you should sit 2m away from the loudspeakers directly in front of them. The centres of the speakers should then be approximately 35cm apart. If your speakers are very small, try to sit 1m away from them when their centres are 20cm apart. If possible, you should position the loudspeakers in the middle of the room; the fewer reflections you get from the acoustic environment the better."

I tried playing the links you posted.

I set up some full-range single drivers with the centers 35 cm apart (like they tell you to do, in the recording) and sat directly in front 2 meters away (as per the instructions).

In the link http://www.isvr.soton.ac.uk/FDAG/VAP/wave/track2.wav,  as the guy says "You should hear my voice coming from well to the left of the loudspeakers", his voice does indeed come from way outside the left speaker.

If you shift your head to the right (about 6 to 8") while he's talking about "hearing my voice well to the left of the loudspeaker", his voice is now coming from the far outside the right loudspeaker.

I've had 4 people do it, and get the same results.

Try it.

Cheers

[Russell Dawkins]

interesting link, Dayglow!

[arthurs]

.....man this thread went quiet.....

DGlow - Was there a track for voice outside the right speaker?  Did the resulting change of shifting your head replicate itself? 

[csero]

Yes, with legacy stereo material and 2 speakers the sweetspot is small, just like with stereo in a dead room. This is just a technology demo, not a full fledged solution.
There are further technologies to increase sweetspot and consistency in case of head movement. Also the extra simulated ambience channels or an ambisonic array can create an environment which in a way similar to real life - focused and detailed if you sit down and listen, much more vague but still believable if you just walk around in the back of the performing venue.
Real homogenous reproduction requires different recording approach than today's 2 ch or MCH methods. Heard a couple of experiments and they are no question superior to this methods, just like is this method is superior in reproducing believable acoustic environment in a room acoustically very different from the recording venue in case if we have only 2 recorded channels and can live with an one man sweetspot.

[sunshinedawg]

Here is some good info on ITD's and ILD's.

[Daygloworange]

.....man this thread went quiet.....

DGlow - Was there a track for voice outside the right speaker?  Did the resulting change of shifting your head replicate itself? 

Uh huh.

And yeah, this thread did get quite after I posted about those links, didn't it?

Yes, with legacy stereo material and 2 speakers the sweetspot is small, just like with stereo in a dead room. This is just a technology demo, not a full fledged solution.

You only get one chance to make a first impression.

And yes, the "technology demo" did impress me.       I've never had an image jump from one extreme end of a L/R image, to the extreme other end of a L/R image simply by shifting my head 6 to 8" off the center axis of a pair of speakers before.

I'd really like to hear from the proponents of transaural, and ambiophonic sound reproduction in this thread ( who have done nothing but bash 2 channel audio, as a feeble attempt at realistic sonic reproduction), who have professed firm understandings of psychoacoustics, to describe, in technical terms, what is happening to the 2 channel signal in the demo, to make it jump from one speaker to the other when you move your head a number of inches off axis?

I'd really like to understand better, what it is, that I (apparently) don't understand about the subject.

As for me, I'm taking this 2 channel demo into the studio with me, and play around with a couple of things with this recording....

Cheers

[Daygloworange]

Here is some good info on ITD's and ILD's.

These are mostly speculations on human perception, and the ear/brain mechanism. Hardly conclusive. Interesting hypothesis's, but speculations nonetheless.

Cheers

[sunshinedawg]

And yeah, this thread did get quite after I posted about those links, didn't it?

Yeah, you set the record straight after that post.   

[opaqueice]

Well, one thing that does seem pretty apparent is that the proponents of this method don't understand it well enough to explain it.  That's not to say it doesn't work - but you might want to tone down the rhetoric a tad.

If I had to hazard a guess as to why the image jumped that way, it would be that the recording was phased in such a way as to cause a null at one ear when you were in the sweet spot.  Suppose the null was at your right ear; then the image would be very far to the left.  Now if you moved your head to the right, your right ear would move out of the null and your left would move in, making the image suddenly jump far to the right.

[Daygloworange]

And yeah, this thread did get quite after I posted about those links, didn't it?

Yeah, you set the record straight after that post.   

Well, your reluctance (some might speculate, inability) to elaborate on what it is, that this demo shows is being done to the signal to cause this anomalie, speaks volumes to me.

opaqueice,

You've earned my respect by stepping up to the plate, and sharing your thoughts on what might be going on. 

I only had a bit of time to import those files into my digital workstation and run a few quick things to see what might be going on, but to avoid any more conflict here, I'll just keep my findings to myself.

Besides, what do I know?   

Cheers

[opaqueice]

I only had a bit of time to import those files into my digital workstation and run a few quick things to see what might be going on, but to avoid any more conflict here, I'll just keep my findings to myself.

Besides, what do I know?   

Well, I at least would be very interested to hear what you found.  I'm curious whether my speculation was correct, for one thing.

[dwk]

I'd really like to hear from the proponents of transaural, and ambiophonic sound reproduction in this thread ( who have done nothing but bash 2 channel audio, as a feeble attempt at realistic sonic reproduction), who have professed firm understandings of psychoacoustics, to describe, in technical terms, what is happening to the 2 channel signal in the demo, to make it jump from one speaker to the other when you move your head a number of inches off axis?

I'd really like to understand better, what it is, that I (apparently) don't understand about the subject.

As for me, I'm taking this 2 channel demo into the studio with me, and play around with a couple of things with this recording....

Cheers

Well, I don't consider myself a 'proponent' of any particular viewpoint in this debate, but I'll take a shot since some of it seems straightforward.

The basic idea behind the front stereo dipole is crosstalk cancellation - emitting an additional signal from the left channel to null out the contribution that the right speaker makes to the left ears response and vice-versa. This means that you need a 180 degree phase shift in the correction signal. The timing of this signal is dependent on the geometry of your ears which are - get this - about 6-8" apart. So what happens when you move your head an additional 6-8" (we'll use left for this)? Basically, you've moved your fight ear into the spot where your left ear used to be; thus what it will hear is actually only the left channel, since the right has been nulled out at this point in space. This won't be exact due to HRTF, but it seems reasonable that it would be close enough. The left ear will be an additional distance to the left, and the effect will be frequency dependent where some cancel an some reinforce - in a range around ~1kHz it'll hear nothing but the right channel since it is an additional 180 degrees away.  So, when the image 'jumps' I would expect that it's not a complete total jump since the left ear will still get some left signal, but it will be nulled out in some critical areas which when coupled with the complete inversion in the right ear may well result in perceptual inversion.

I believe that this is part of thrust behind an enhanced technique that someone referenced earlier, proposed in a paper titled something along the lines of 'Optimal source distribution for x-talk cancellation' where they show that the angular separation of the speakers should vary with frequency to provide stable cancellation. Tweeters in front, mids out ~30 degrees or so (depends on the range covered), bass drivers to the side (theoretically it should be a continuous function, but that's not practical). The main point of this is that such an arrangement allows the use of a 'simple' 90-degree phase shift in the cancellation signals, but it also provides better stability - as much as can be expected in an inherently position-sensitive approach.  I've always wanted to try this since it should be relatively straightforward using BruteFIR which I already have setup; never found the time, though. (I also suspect that a frequency-dependent windowing component is likely necessary, since at treble wavelengths cancellation is going to be insufficiently stable regardless of the close driver spacing)

FWIW I do find you continued insistance that the equilateral setup 'properly' inverts the recording process baffling. Two speakers clearly and irrefutably create 4 acoustic arrivals in the direct field where originally there were only 2. There certainly is lots of room to debate how much this matters and whether any particular xtalk cancellation scheme is better, but to deny that there is a fundamental problem is groundless. I think that particularly in light of Geddes work showing that diffraction and early reflections are a) more perceptually intrusive than originally thought and b) level dependent, there's room to re-evaluate the fundamental assumptions of stereo, due to the similarity of the delayed xtalk signal to a diffraction effect.
 That said, I use a setup that is pretty much equilateral and quite like it, so saying something is theoretically flawed is not the same as saying it doesn't work acceptably.

[opaqueice]

FWIW I do find you continued insistance that the equilateral setup 'properly' inverts the recording process baffling. Two speakers clearly and irrefutably create 4 acoustic arrivals in the direct field where originally there were only 2.

I don't think that's a good way to think about it.  You've got to bear in mind that we're not talking about narrow beams here: we're talking about waves, and two waves sources make a pattern of constructive and destructive interference.  So I don't think it makes sense to talk about "4 arrivals" - it really doesn't mean much.  There's just one value for the air pressure at a given place at a given time, and it's the sum of all the different sound waves impinging on that point at that moment, end of story.

If you really want to hear something very close to what a listener in the venue would hear, you take a dummy head, put it in a chair, put mics in each ear, and record.  Then play that back with phones.  No speaker system is ever going to beat that - or even get close.

[Daygloworange]

FWIW I do find you continued insistance that the equilateral setup 'properly' inverts the recording process baffling. Two speakers clearly and irrefutably create 4 acoustic arrivals in the direct field where originally there were only 2. There certainly is lots of room to debate how much this matters and whether any particular xtalk cancellation scheme is better, but to deny that there is a fundamental problem is groundless. I think that particularly in light of Geddes work showing that diffraction and early reflections are a) more perceptually intrusive than originally thought and b) level dependent, there's room to re-evaluate the fundamental assumptions of stereo, due to the similarity of the delayed xtalk signal to a diffraction effect.
 That said, I use a setup that is pretty much equilateral and quite like it, so saying something is theoretically flawed is not the same as saying it doesn't work acceptably.

dwk,

I have neither the desire, nor inclination to debate any of this any further.

This thread is full of people making too many declarations of absolutes, which of course is to fend off any possibility of opposing opinion.

I have no idea what you mean when you are talking about 4 acoustic arrivals emanating from a pair of speakers to the listener. I'm not even interested enough to ask you to elaborate. You and I would probably be on opposite ends of a debate, which will do nothing but turn into a pissing contest.

All this "new" technology being presented here, is to try and duplicate a 360 deg soundfield.

It is trying to do multi channel by taking 2 channels and synthesizing the remaining surround channels from those, based on DSP technology. Knowing what I know about DSP technology, and many other things about sound, I find it silly to do it this way. There's a better, more linear way to skin that cat. It was invented almost 40 years ago.

Just for the record,

I've been there, done that, many years ago. Back in the 80's, I got heavily into the Q4 based Quadraphonic technique, which is 4 discrete analog tracks (on reel to reel) fed to (2) independant 2 channel systems.

I created my own recordings in my studio, where I had access to a large ambient space that was a large hall (with wood floor,  wood panelled ceilings, and walls) and had spent literally thousands of hours, experimenting with different mic techniques (X-Y, spaced pair, baffled spaced pair, Blumlein, binaural ) and many different iterations of qaud speaker placement, including separating speakers by (treated) center dividers.

So I have a lot of theoretical and emperical knowledge on the subject. I've also forgotten more about the mathematical theory than I remember, as I've moved on to other things, and I just don't care enough to convert anyone to my thinking.

Also just for the record, I'm a music lover first, a classically trained musician second, a recording engineer third,an audiophile fourth, and most recently, a manufacturer of speakers.

Quite frankly, based on how easy it is to get into heated arguements with audiophiles, I care very little about the fourth 

So, I guess I'll just have to agree to disagree with (apparently) many audiophiles, and just do my own thing. 

Cheers

[dwk]

FWIW I do find you continued insistance that the equilateral setup 'properly' inverts the recording process baffling. Two speakers clearly and irrefutably create 4 acoustic arrivals in the direct field where originally there were only 2.

I don't think that's a good way to think about it.  You've got to bear in mind that we're not talking about narrow beams here: we're talking about waves, and two waves sources make a pattern of constructive and destructive interference.  So I don't think it makes sense to talk about "4 arrivals" - it really doesn't mean much.  There's just one value for the air pressure at a given place at a given time, and it's the sum of all the different sound waves impinging on that point at that moment, end of story.

You're missing the fact that the properties of the 'single acoustic field' are very very different in the presence of two sources than with only one.

Basically, you end up with a matrix problem:

  [L(t)    LR (t-td)] Xl  =   [I(x,t), I(r, t)]
  [RL(t-td)   R(t)  ] Xr   

With two sources, the off-diagonal elements are non-zero, whereas for a single source they are zero. The entire problem of crosstalk cancellation is to create an additional matrix of filters that when multiplied by the original matrix results in the 2x2 identity matrix.  This is why the whole problem is called "2x2 crosstalk cancellation".

So, once again it's possible to debate how important it is, but mathmatically it's not possible to debate that it's real.

If you really want to hear something very close to what a listener in the venue would hear, you take a dummy head, put it in a chair, put mics in each ear, and record.  Then play that back with phones.  No speaker system is ever going to beat that - or even get close.

Sure, but dummy head recordings are rare, and not everyone has the same HRTF. I do wish there was more work done in this direction, though.

[sunshinedawg]

FWIW I do find you continued insistance that the equilateral setup 'properly' inverts the recording process baffling. Two speakers clearly and irrefutably create 4 acoustic arrivals in the direct field where originally there were only 2. There certainly is lots of room to debate how much this matters and whether any particular xtalk cancellation scheme is better, but to deny that there is a fundamental problem is groundless. I think that particularly in light of Geddes work showing that diffraction and early reflections are a) more perceptually intrusive than originally thought and b) level dependent, there's room to re-evaluate the fundamental assumptions of stereo, due to the similarity of the delayed xtalk signal to a diffraction effect.
 That said, I use a setup that is pretty much equilateral and quite like it, so saying something is theoretically flawed is not the same as saying it doesn't work acceptably.

Yes, thank you. I know sometimes I don't explain things in the best way, but like dwk said you can't deny the problems. They are basic physical laws, one thing can't be in different places at the same time and also something can't be in one place and appear to be somewhere else. Further, a triangle setup is not going to reverse an unbreakable law of physics. This is not really arguable. If you are willing to accept these flaws that is your prerogative. 

[doug s.]

i have found this conwersation quite interesting.  sitting on the fence as a casual obserwer, as it were.  i used to own a jvc xpa1010 dsp, which allowed the set-up of four ambient speakers to mimic 20 different venues from real locations which were programmed into the dsp - churches, concert halls, clubs, etc.  with some software, the effect was eerily realistic.  that said, i rarely used it.  why?  because, in most instances it was not really better than my normal 2-channel playback system, yust different.  in spite of some statements in this thread that were made as facts, about what is not physically capable of being done with 2-channel audio, i easily experience a soundstage that's detached from the speakers, wider than the speakers, higher than the speakers, with real depth of the soundstage, etc.  this is no anomaly; it's present in many recordings, and on a lot of the fm i listen to.  so, while it may be impossible, i experience it on a regular basis.   

also, while i understand that when a pair of stereo speakers is playing, there are four distinct arrivals - both your ears hear each speaker.  but, i also know that when i listen to a live duet on a stage, there are also these same four distinct arrivals.  yust cuz the duet is live, doesn't mean my left ear isn't hearing the right player & wice-wersa.  perhaps this helps to explain why conwentional two-channel audio playback can do such a credible yob of reproducing a believable soundstage with decent recordings...  even when hearing a believable soundstage from two-channel audio is flatly impossible, according to some of what i have read here...   

ymmv,

doug s.

[sunshinedawg]

I have neither the desire, nor inclination to debate any of this any further.

This thread is full of people making too many declarations of absolutes, which of course is to fend off any possibility of opposing opinion.

I hope you change your mind, you have already offered a lot to this thread. I am enjoying this thread, I have already learned stuff I didn't know.

You said earlier:

"It can't. I have many issues with how typical (read: popular) 2 channel recordings are made. I wish 2 channel audio was 5 channel audio from day one."

Do you have anymore to offer on this statement? How would you implement 5 channels?

The only absolutes I meant to say or imply where the ones about the fundamental laws of physics. I hope I didn't give the wrong impression about anything else.

I'm glad that people are thinking about setups. I wasn't sure if anybody had considered the science behind it because I had come to such a different conclusion. I especially enjoy the conversation about how people perceive different system setups.

The other thing that has got lost in this thread is how good I think the CS2's are. I did make my decision based on hearing them in a triangle setup. I never heard a speaker that can do what this one does, especially with no room treatment. $3000 is a lot of money to me. Probably equal to what I paid for the rest of my system. I think that speaks volumes to what I think of them