A theoretical model for stereo imaging

[geickmei]

Keith -

So how did you test any of that?

No, we do not want a reverberant field in our listening rooms that exceeds the recorded one, but there are some interesting effects that happen as you do that. Once upon a time I went to a stage magic show. The performer used two stacks of Bose 802s, one on each side in this huge room. I found that once the actual acoustic overwhelms that in the recording it sounds so real it cannot be distinguished from a real performance. It's when we get into smaller rooms that the problem becomes more difficult - trying to duplicate the acoustics of a larger recording room in a smaller listening room. You need to attempt to get all of the spatial, spectral, and temporal characteristics to be about the same as they were in the spot where the mythical mikes were located, plus or minus. Recorded reverberant sounds must come from wide incident angles and delayed enough to separate them from the direct field but not so much as to exceed the fusion time. Spectral must be equalized for the power response, not the axial from the speakers. Direct sound should come directly from the speakers, positioned in about the stereo triangle - out from all walls and simulating the approximate positioning of the musical instruments. Surround sound, real or upmixed, should blend with that soundstage and bring the ambience farther around to complete the spatial picture.

More channels equal greater realism if you know what to do with them. I record in MS and play in Dolby Surround or similar for a near discrete surround sound, with the audience all around me and not coming from behind the orchestra.

In my experience this kind of image modeling works for any kind of performance, live or multitracked. If it is live, all of the spatial elements fall into place - classical, jazz, some pop. You are there. If it is tracked, they sound like they are right in your room with you. They are here.

I haven't got it perfect yet, but I'm working on it. A large company like Harman should be able to perfect Image Modeling in a couple of weeks...

Gary

 

[dlaloum]

Keith -

So how did you test any of that?

No, we do not want a reverberant field in our listening rooms that exceeds the recorded one, but there are some interesting effects that happen as you do that. Once upon a time I went to a stage magic show. The performer used two stacks of Bose 802s, one on each side in this huge room. I found that once the actual acoustic overwhelms that in the recording it sounds so real it cannot be distinguished from a real performance. It's when we get into smaller rooms that the problem becomes more difficult - trying to duplicate the acoustics of a larger recording room in a smaller listening room. You need to attempt to get all of the spatial, spectral, and temporal characteristics to be about the same as they were in the spot where the mythical mikes were located, plus or minus. Recorded reverberant sounds must come from wide incident angles and delayed enough to separate them from the direct field but not so much as to exceed the fusion time. Spectral must be equalized for the power response, not the axial from the speakers. Direct sound should come directly from the speakers, positioned in about the stereo triangle - out from all walls and simulating the approximate positioning of the musical instruments. Surround sound, real or upmixed, should blend with that soundstage and bring the ambience farther around to complete the spatial picture.

More channels equal greater realism if you know what to do with them. I record in MS and play in Dolby Surround or similar for a near discrete surround sound, with the audience all around me and not coming from behind the orchestra.

In my experience this kind of image modeling works for any kind of performance, live or multitracked. If it is live, all of the spatial elements fall into place - classical, jazz, some pop. You are there. If it is tracked, they sound like they are right in your room with you. They are here.

I haven't got it perfect yet, but I'm working on it. A large company like Harman should be able to perfect Image Modeling in a couple of weeks...

Gary

In the mid 1980's Yamaha went and analyzed the "signatures" of a bunch of famous performance venues all over the world (and then built models of them into their surround processor)...

Would that work be of any interest / relevance to this discussion?

 

[Keith_W]

Keith -

So how did you test any of that?

Hi Gary. As mentioned in that link, a "reverberant field" should measure the same no matter where the measurement is taken in the room. To test whether I have reverberant fields in my room, I took RT60 measurements from these positions:

I then overlaid all of them and looked for divergence. If there is no divergence, or minimal divergence, then it is a reverberant field:

In my room, the end of the transition zone (4x Schroder) is 532Hz, which is about where the measurements start to diverge. Divergence indicates that the RT60 decay measurement is dependent on microphone position, meaning it is measuring room modes rather than a reverberant field.

I don't know if this is a known technique or not, it is something I came up with. The reasoning seems sound to me, but then I have never had a real acoustics expert look at my technique and criticize it.

 

[goat76]

Goat, if you would please read my article you could see how I cover this duality of methods or systems. Jens Blauert states it very precisely as field type vs head related, reconstructing a "synthetic sound field" modeled after the real thing vs binaural. Stereo played without cancelling the crosstalk can be very good but the sound is limited to between the speakers. I think what you are thinking of is loudspeaker binaural, which is not stereo. The larger discussion is in my article.

Gary

I have not read your article, but based on the thing you said, which was the following: "CERTAINLY live sound does not come from two points in space spouting out a narrowly dispersed stream of direct sound toward your ears.", it seems to me that you are a little bit confused about what the task of the loudspeakers is, and what kind of information a stereo recording can contain.

Still, it's not the job of the loudspeakers to "mimic" the sound dispersion characteristics of the live sound field, as that information is already in the stereo recording itself (granted the ambition to do so was there to begin with). That information can contain most of the acoustic qualities of the live sound, including most of the spatial directional cues.
We, as listeners, are not listening to "two points in space"; we are listening to the whole stereo image, which contains multiple directional cues of both the direct sound of the recorded instruments and the directional cues of the acoustic live sound field and reflections. All those sounds come with the direct sound from the loudspeakers.

Anyway, I'm referring to ordinary stereo audio productions. It's not uncommon to widen the stereo field or enhance individual elements with some out-of-phase trickery, as in the following examples, which I just created a few minutes ago.

Hard panned drums and guitar: https://www.dropbox.com/scl/fi/a35i...ey=8dth5mmbycuiaxsa48syrk23t&st=effd2xrr&dl=0

Hard panned drums and guitar with extra widening a bit outside each loudspeaker: https://www.dropbox.com/scl/fi/p3v9...ey=1a34xpwj0csc9p9yjxxwaeeon&st=wrr4mlx1&dl=0

 

[kawauso]

Hi Gary. As mentioned in that link, a "reverberant field" should measure the same no matter where the measurement is taken in the room. To test whether I have reverberant fields in my room, I took RT60 measurements from these positions:

View attachment 491378

I then overlaid all of them and looked for divergence. If there is no divergence, or minimal divergence, then it is a reverberant field:

View attachment 491379

In my room, the end of the transition zone (4x Schroder) is 532Hz, which is about where the measurements start to diverge. Divergence indicates that the RT60 decay measurement is dependent on microphone position, meaning it is measuring room modes rather than a reverberant field.

I don't know if this is a known technique or not, it is something I came up with. The reasoning seems sound to me, but then I have never had a real acoustics expert look at my technique and criticize it.

RT60 is sometimes dismissed in small rooms because they aren’t true reverberant fields, but your approach makes it clear that the mid and high frequencies can still be meaningful. It’s also a very intuitive metric to look at, so I want to keep using it.

The data under about 500 Hz might not reflect a real decay time, but in your graphs it still looks surprisingly well-behaved. Is your room treated for low-frequency absorption? The consistency down there is unusual in a good way.

Also, I never associated you with horn speakers, so seeing that system was a surprise. It looks fantastic.

 

[geickmei]

In the mid 1980's Yamaha went and analyzed the "signatures" of a bunch of famous performance venues all over the world (and then built models of them into their surround processor)...

Would that work be of any interest / relevance to this discussion?

Dlalum -

No, not for me. I don't like effects processors. Rather have the real acoustics recorded.

And oops - I told the story wrong. The stage magician had the two stacks of 802s on opposite sides of the stage, not the room. So that was like having the band onstage with him, the music stretching the same space as a live band would have within that performance space. I did the same trick with a new pair of Magneplanar MG-30s in a huge room that they had for a demo. In a case like that you don't try to make the sound as wide as a movie screen (if there is no movie), you make it as wide as a real group would be if they were present. They are here!

Gary

 

[geickmei]

I'm getting confused as to who is speaking here - I never showed any graphs of my reverberant field and I don't have horns.

Gary