I believe it has everything to do with the room, more precisely the reflection pattern in the room. It's "simply" the lack of distorted spatial cues that makes our hearing believe the auditory scene.
It's quite easy actually. All is needed is to sufficiently block the direct sound from the speaker. The speaker position is very hard to be determined based on reflections only.
I just found out this thread has turned into something quite interesting, so I backpeddled
some 10 pages or so and I feel this remark by Barleywater hits the nail on the head.
Some further reflections:
- The Technical Universtity of Delft did some experiments decades ago to improve stereo imaging by having a horizontal array of loudspeakers instead of the ordinary two. But, obviously, to know what to feed into that array, you first need measurements in room of live performances/ measurement signals in order to determine what you want to hear back. This is where the problem occured. Even small differences in position in the room would lead to significant modifications of the sound profile, both in FR and arrival time. This effect is recursive in the sense that variations on a large scale are comparable to those on much smaller scales. We seem to localize sound by navigating through all these combfilter and reflection effects by minute head movements. Needless to say that the horizontal array of speakers never took off.
- With loudspeaker localization, it is exactly as Barleywater describes. The only way around is to make speakers that have the smallest possible diffraction signature. The effect heard by Pano with the horns with backfolded mouth opening is a case in point. The reason for a backfolded mouth is exactly to prevent diffraction products being reflected back into the horn, producing distortion.
- In general I think there are two more or less equivalent but opposite technologies that can produce the effect mentioned by Pano. The first is to have a vertical and horizontal dispersion window in which FR deviations are well controlled. A good horn system or dipole does this. Only downside is that of beaming, and the necessity to choose the listening position exactly at the location where the beams cross. The other solution is to create loudspeakers that act as a point source, like the MBL's. This way, it is possible to have a very wide sweet spot, basically filling most of the room. Linkwitz's Pluto goes some way towards this.
That was also my experience. Always two speakers playing, IIRC, stereo source. In the examples I remember it was not omni directional. The effect would collapse somewhere behind the speakers, tho I don't remember at what point.
With my modified VOTT A5 speakers, I can walk right up to the box, almost touching it - the sound seems to come from just behind the speaker, not from the horn or the woofer. It's a strange effect.
I agree with this, except that I would be careful calling "diffraction" "distortion". Diffraction is a completely linear process. Its audibility is nonlinear however.
Having just heard the MBL speakers yesterday and they had a smooth response, but no imaging at all. The Omni response just had so many early reflections that no imaging was evident at all.
The new Linkwitz speaker was better at imaging.
A few points from what I am reading here. The original statement if I remember correctly was that by walking around the speaker you could not localize where the sound was coming from. Not that you had a disperse sound field but that you could not identify where the sound source was. The second thing that was one of the statements was if you block the front of the speaker so that it is hidden and that was not stated in the least.
An omnidirectional speaker may come close to that but most actual implementations do have a problem in the vertical direction. The MBL is as close to what I originally said that it would have to be a pulsating sphere and that implementation is about as close to that ideal as I have seen done, so that goes with what I was saying.
What I now believe that Pano is saying is that the apparent center is behind the speakers the acoustical center does not seem to be coming from the driver or the horn. These are all different room effects that are hard to reproduce without an exact positioning of the speakers and mostly luck with the dimensions of the room and the reflectivity and dispersion pattern. Once this happens you would never want to move the devices as as soon as you do that the imaging would be lost by a change in pattern and interaction would occur. Now what I was saying and that is the hard part to get this to ever happen is that you can walk up to the speakers and not identify any localization at any frequency. That would be very outside the norm.
Now if you could do this in a predictable way that would be a real accomplishment, not something that most waveguides or direct radiators is going to do, perfect integration with the room. I was helping one of the members in a separate thread who is trying to build a diy MBL clone, last I spoke with him he had procured the voicecoils and is attempting to build the rest of the speaker, thee leaves of the MBL will be difficult to reproduce without knowing all the resonant properties of the originals and the ways in which they dampened those leaves. If I hear more about how the build is going I will let you know.
An omnidirectional speaker may come close to that but most actual implementations do have a problem in the vertical direction. The MBL is as close to what I originally said that it would have to be a pulsating sphere and that implementation is about as close to that ideal as I have seen done, so that goes with what I was saying.
What I now believe that Pano is saying is that the apparent center is behind the speakers the acoustical center does not seem to be coming from the driver or the horn. These are all different room effects that are hard to reproduce without an exact positioning of the speakers and mostly luck with the dimensions of the room and the reflectivity and dispersion pattern. Once this happens you would never want to move the devices as as soon as you do that the imaging would be lost by a change in pattern and interaction would occur. Now what I was saying and that is the hard part to get this to ever happen is that you can walk up to the speakers and not identify any localization at any frequency. That would be very outside the norm.
Now if you could do this in a predictable way that would be a real accomplishment, not something that most waveguides or direct radiators is going to do, perfect integration with the room. I was helping one of the members in a separate thread who is trying to build a diy MBL clone, last I spoke with him he had procured the voicecoils and is attempting to build the rest of the speaker, thee leaves of the MBL will be difficult to reproduce without knowing all the resonant properties of the originals and the ways in which they dampened those leaves. If I hear more about how the build is going I will let you know.
It is my experience that with 1 speaker playing, it is best when you can pinpoint exactly where the speaker is. Phase issues will reveal themselves by creating faux displacement effects when playing mono.
With 2 speakers playing, it is best when you can no longer pinpoint any of the two speakers at all. This happens when the wavefronts from the two speakers seamlessly combine.
With 2 speakers playing, it is best when you can no longer pinpoint any of the two speakers at all. This happens when the wavefronts from the two speakers seamlessly combine.
Sorry if I seemed to say that. I did not mean "all around", more like 200 degrees, or 100 off the center of the speaker face. I don't remember exactly.
Correct, the sound appears to be somewhere else, not at the speaker. In the case of my speakers, it seems to come from behind them.
Could be, but I don't think so - as at least two of us have heard in very different rooms. A good room can't hurt, I'm sure.
Yes, I agree. It is outside the norm, very few systems can do it. You don't forget it when you hear it. Jean Hiraga and the crew at l'Audiophile Paris seemed to be able to pull it off with regularity, so it's that approach that I follow. That's were I first heard it.
I am not sure what "prefect integration with the room" means, but I don't see why an Omni would be better at that than a waveguide.
I am not up to the current conversation, but this sounds like the speaker "disappears". This is the first thing that I listen for when I listen to unfamiliar speakers. Any speaker that fails this first test, I don't go to a second step. The second step is, does a solid image on the recording, remain a solid image on playback? Not all recordings have this, of course, I know many that do, so its not easy to implement this test at say a show. The fact is only a few speakers at AXPONA passed the first test. None of those had source material that I knew had the imaging that I am looking for.
Earl,
I guess what I mean in that statement is that the diffuse sound field makes isolation of the speaker from the sound field hard to hear, as your number 1 requirement you just listed. Even a speaker that is good at this would have to be very precisely places at different locations every time it is placed in a room with different size, reflectivity and room modes. You can't just take a speaker that sounds good in one room and drop it willy nilly into another room and expect this to happen. That is the problem with most demonstrations at most shows, the dimensions of the room can just be such and the acoustical properties of the room just make it next to impossible to achieve that audio nirvana. If you had days to get it right perhaps, but that has never been the case when I did the CES or Stereophile shows, it was a quick entry into the room you have never seen before and setup, not very conducive to great audio I must say. Sometimes you have to forgive the people who are showing a pair of speakers for not being able to optimize a room, but if you listen carefully you can at least hear if the proposition is possible. If the sound balance is so awful or the dynamics are just not there I just move on. Sometimes you don't even have to enter the room, you know immediately that the speaker has no dynamic quality or the balance is so far off you just keep walking.
I guess what I mean in that statement is that the diffuse sound field makes isolation of the speaker from the sound field hard to hear, as your number 1 requirement you just listed. Even a speaker that is good at this would have to be very precisely places at different locations every time it is placed in a room with different size, reflectivity and room modes. You can't just take a speaker that sounds good in one room and drop it willy nilly into another room and expect this to happen. That is the problem with most demonstrations at most shows, the dimensions of the room can just be such and the acoustical properties of the room just make it next to impossible to achieve that audio nirvana. If you had days to get it right perhaps, but that has never been the case when I did the CES or Stereophile shows, it was a quick entry into the room you have never seen before and setup, not very conducive to great audio I must say. Sometimes you have to forgive the people who are showing a pair of speakers for not being able to optimize a room, but if you listen carefully you can at least hear if the proposition is possible. If the sound balance is so awful or the dynamics are just not there I just move on. Sometimes you don't even have to enter the room, you know immediately that the speaker has no dynamic quality or the balance is so far off you just keep walking.
(I fixed your previous post)
It's the big blocks of text that are hard to read on a computer screen.
I agree. Omnis in a bounded space simply don't work if one desires anything close to accuracy. The sound is everywhere and nowhere. MBL gave me maybe one of the worse listening experience I've ever had. I would rather buy a cheap PSB speaker.