What exactly is "imaging" ?

[tuga]

Have you never listened to the test track I mentioned a few days back? Nobody responded to that post.

Which track, the rainbow arc? There's one here: https://www.audiocheck.net/audiotests_ledr.php

OVER. The sound should begin at one speaker and travel in a smooth arc to the other speaker, from left to right and then return back to the left. The arc should be unbroken, smooth and symmetrical. The top of the rainbow should be as high as the Up signals.

It is possible to manipulate phase in order to place the phantom image outside of the space between the two speakers.
I don't find either the UP effect or the OVER effect of source describing a rainbow arc very convincing illusions with speakers, they work better for me with headphones.
But I am partly deaf and my system is cråp...

 

[Sue Pertwee-Tyr]

Yes, the over effect, that’s the one. I find it very convincing. And right at the start of this thread, I referred to phase as a key element in imaging, so why should it not provide cues for height, too?

 

[adamdea]

Re: what I quoted from Linkwitz earlier. Paraphrase, our hearing systems are extremely good at locating the 3D position of sound sources in the presence of other objects (e.g. the ground, walls etc.)

Putting speakers in a room, our aim is to fool the brain exactly in the area it's phenomenally good at. We DON'T want our brains to locate the real sound source(s) at all. We want the brain to perceive phantom sound sources.

Linkwitz argued that it's possible to achieve it in a room, by working with reflected sound, rather than against it, and having an even frequency response at every angle off-axis. Then the brain stops perceiving the speakers and even the surfaces in the room, thereby revealing the recording.

If you're saying the perception of the original acoustic will be distorted, degraded and reduced with a non-binaural/HRTF set up, I agree. However, you're saying no height information can be encoded at all because there are just two sound sources, which I think is a misunderstanding, since binaural can encode 3D volumes with two sound sources. What you're really arguing about is the absolute necessity of HRTF/pinnae processing so without this being handled, 3D information is completely absent - and I don't see that as necessarily and absolutely true. I don't see why some information can't be present (I've repeatedly said that it might not be technically accurate, just that something is present).

Darren
Yes I am arguing that without some sort of HRTF pinnae processing an ordinary stereo mic set up cannot encode height information (which is not the same as whether it can record the sound of things coming from the top; it's how it differentiates them from things not coming from the top). [Let's not confuse this with "3D" which is an ambiguous expression (see below)]. I honestly believe this to be the orthodox view but woud be interested to hear otherwise how height (angle of incidence) information is encoded in practice, if it is.
[I wish i had stopped here, but I feel obliged to show my working.]

What this comes down to is that as I understad it in the real worlld you detect/decode height information (ie information as to the angle from which sound comes) using pinna/head spectral cues and by head moving (and of course using non aural cues) not by interaural time and amplitude differences. If I have missed some other auditory information as to the angle of incidence of sound then what is it? ie what is the "height information". There is no equivalence between sterophonic hearing and stereoscopic vision. You don't get the equivalent of an image in a particular locaiton (mapping to external spatial location) on your retina.This is crucial.

Without those spectral cues being encoded into the signal you don't get the height information (AKA "comes from above your head" information). This is presumably why we don;t hear the orchestra coming from below the speakers or get a sense of looking down at the orchestra.

And even if the spectral information were encoded and replayed by stero speakers, it would then be confused by the fact that the sound was coming from the front and will all be impressed wiht your head's "sound in front" spectral imprint, so it will only work with headphones. The other way to do it is to actually encode separate "above" channel info (no spectral cues required) and send it to your head from the right direction. Your head then supplies the comes "from the top" spectral imprint. The fact that reflections from the ceiling will be recorded does not tell you how it is encoded specfically as being refections from roof not reflections from the wall. That is the nub of the matter, as is perfectly clear if one starts from the question of how we perceive height information.

The fact that your brain may create/interpolate/guess/imagine that some sound comes from above or below sometimes does not mean that height information is encoded. If you imagine a room you tend to imagine one which has some sort of height dimension. If you imagine a violin you tend to imagine one which has some sort of size. It would be surprising if your brain did not fill in some information.

And while we are on it. I have not notificed (on an admitedly quick reading) anyhting in the linkwitz article on height information as such. I think you are reading far too much into his use of the expression "3D" in one sentence. Can you point me to any passage in which he refers to the encoding of height informaiton explictly?

I have not noticed anything on the net about how to encode/ master in height information in stereo (there's plenty on l-R panning and depth). If it can be done I would be interested to read it.

If anytone thinks this is just theorectical then I'm all ears about how sound engineers set out to record (so as to be diffentiated by listeners) instruments at different heights in the sound field or even better vertical panning. I have noticed that in terms of left right panning some panning software does use some sort of left /right spectral processing as well as time and amplitude differences. Doies such a vertical panning device exist for stereo?

 

[adamdea]

Which track, the rainbow arc? There's one here: https://www.audiocheck.net/audiotests_ledr.php

OVER. The sound should begin at one speaker and travel in a smooth arc to the other speaker, from left to right and then return back to the left. The arc should be unbroken, smooth and symmetrical. The top of the rainbow should be as high as the Up signals.

It is possible to manipulate phase in order to place the phantom image outside of the space between the two speakers.
I don't find either the UP effect or the OVER effect of source describing a rainbow arc very convincing illusions with speakers, they work better for me with headphones.
But I am partly deaf and my system is cråp...

Quoting from that very page
"EDR™ stands for Listening Environment Diagnostic Recording, a test to subjectively evaluate the accuracy of stereo image reproduction.

In the eighties, psychoacousticians began researching what are called pinna transforms, the way in which the shape of the outer ear filters the incoming sounds and permits our brain to infer their location. By embedding the filtering characteristics of the pinna into the audio signal, sound can be moved around the listener's head from a single pair of loudspeakers.

The LEDR test generates pinna-filtered audio that will literally float around your speakers, assuming your sound reproduction system is neutral enough to preserve the original signal characteristics."

Hello

 

[tuga]

Yes, the over effect, that’s the one. I find it very convincing. And right at the start of this thread, I referred to phase as a key element in imaging, so why should it not provide cues for height, too?

It is a manipulation of the signal so not possible with real stereo live to 2-mic > 2-channel > 2-speakers, only with studio mixes.
As we can hear in the tests I linked, by manipulating phase it is possible to place a mono close-mic'ed source to the outside of the space between two speakers in the horizontal plane, or with in my opinion little success somewhere UP above a speaker or OVER the rainbow arc between them, and with no success BEHIND the listener.

 

[Sue Pertwee-Tyr]

Just to say that, by and large I think any height differences in an image will probably be within +- 15 degrees of horizontal, as perceived from a typical listening position, so it’s not like we’re looking for a huge variance here.

 

[Sue Pertwee-Tyr]

It is a manipulation of the signal so not possible with real stereo live to 2-mic > 2-channel > 2-speakers, only with studio mixes.
As we can hear in the tests I linked, by manipulating phase it is possible to place a mono close-mic'ed source to the outside of the space between two speakers in the horizontal plane, or with in my opinion little success somewhere UP above a speaker or OVER the rainbow arc between them, and with no success BEHIND the listener.

Not many pages ago, you argued that it wasn’t possible due to the dispersion characteristics of the loudspeakers. My point then was that here is a test which proves it clearly is possible, so your argument didn’t hold.
In a real world acoustic, you don’t need to manipulate phase to get the height, the relevant phase relationships would appear already to exist. Provided these are captured by the mic, and preserved during replay, they should help the brain to add height to the image it creates. That’s all I’m saying here.

 

[tuga]

Not many pages ago, you argued that it wasn’t possible due to the dispersion characteristics of the loudspeakers. My point then was that here is a test which proves it clearly is possible, so your argument didn’t hold.
In a real world acoustic, you don’t need to manipulate phase to get the height, the relevant phase relationships would appear already to exist. Provided these are captured by the mic, and preserved during replay, they should help the brain to add height to the image it creates. That’s all I’m saying here.

Can we first decide whether you wish to discuss captured live stereo sound (the "real world acoustic") or manipulated close-mic'ed mono?

 

[Tony L]

Can we first decide whether you wish to discuss captured live stereo sound (the "real world acoustic") or manipulated close-mic'ed mono?

I can’t answer for Steve, but both are valid IMO. If you want to deep-dive this and actually learn stuff have a look at the math that goes into producing the algorithms in digital stereo studio spacial effects such as reverb etc. The behaviour of sound in acoustic spaces and the way humans interpret it is largely understood (though still improving all the time) and the reflections and spacial information can be recreated very convincingly. Modern spacial FX units are very cleverly designed things. A lot of research exists in this field.

 

[adamdea]

Not many pages ago, you argued that it wasn’t possible due to the dispersion characteristics of the loudspeakers. My point then was that here is a test which proves it clearly is possible, so your argument didn’t hold.
In a real world acoustic, you don’t need to manipulate phase to get the height, the relevant phase relationships would appear already to exist. Provided these are captured by the mic, and preserved during replay, they should help the brain to add height to the image it creates. That’s all I’m saying here.

Except, no- that that isn't how it works. Your auditory sytem does not perceive phase differences which already exist in the signa;l. At least they don't exist until the sound hits your pinna and head. So a mic can;t capture it without a dummy head (real or added by post recording manipulation). That's the point.

 

[AndyU]

Stereo can encode some cues for left-right, principally by using amplitude differences. Thus you can have a balance control which will move a mono image one way or the other. There is however no “height” control that can move an image up or down, either during recording or playback. Anyone care to produce one, and I will eat my words. Stereo cannot encode height. Unless of course you rotated a crossed pair through 90 degrees when recording, and played back through speakers aligned vertically. But then ..

 

[tuga]

The behaviour of sound in acoustic spaces and the way humans interpret it is largely understood (though still improving all the time) and the reflections and spacial information can be recreated very convincingly.

The illusion of listening musicians/instruments playing in a reverberant space is reasonably good but the original soundfield can't be recreated becase with 2-channel stereo all sounds (direct, reflected, reverberation) are coming from one or both speakers in front of the listener, not from their original/real direction.

 

[Tony L]

The illusion of listening musicians/instruments playing in a reverberant space is reasonably good but the original soundfield can't be recreated becase with 2-channel stereo all sounds (direct, reflected, reverberation) are coming from one or both speakers in front of the listener, not from their original/real direction.

You really do need to stop speaking in absolutes! To quote Richard Feynman “I’d rather have questions that can’t be answered than answers that can’t be questioned”.

As I say, do a little research into the algorithms in high-end digital reverb and other studio spacial effects. They are often modelling real spaces.

 

[tuga]

You really do need to stop speaking in absolutes! To quote Richard Feynman “I’d rather have questions that can’t be answered than answers that can’t be questioned”.

As I say, do a little research into the algorithms in high-end digital reverb and other studio spacial effects. They are often modelling real spaces.

I was referring to live sound capture not synthesised space effects.

 

[camverton]

The illusion of listening musicians/instruments playing in a reverberant space is reasonably good but the original soundfield can't be recreated becase with 2-channel stereo all sounds (direct, reflected, reverberation) are coming from one or both speakers in front of the listener, not from their original/real direction.

I take it you still haven’t listened to a well set up pair of decent omni speakers. If you had you would know that with them sound appears to come from behind the speakers. This may, in part, account for the realistic illusion of the original performance with these speakers, to the extent that talk of room distortion becomes irrelevant. Stereo reproduction is an illusion, but given the right combination of room, speaker, recording and mind an incredibly convincing one. Basically, your diagram is, at least in my system, fallacious.

No books were burped in the creation of this post; just listening to music and finding out which solution does it best for me with my mind.

 

[tuga]

I take it you still haven’t listened to a well set up pair of decent omni speakers. If you had you would know that with them sound appears to come from behind the speakers. This may, in part, account for the realistic illusion of the original performance with these speakers, to the extent that talk of room distortion becomes irrelevant. Stereo reproduction is an illusion, but given the right combination of room, speaker, recording and mind an incredibly convincing one. Basically, your diagram is, at least in my system, fallacious.

My speakers must be rubbish because I am listening to the Singapore Symphony Orchestra and the sound appears to come from behind the speakers...

No books were burped in the creation of this post; just listening to music and finding out which solution does it best for me with my mind.

How sweet. Who needs books when we have your opinion.

 

[tuga]

An interesting read...

 

[darrenyeats]

And while we are on it. I have not notificed (on an admitedly quick reading) anyhting in the linkwitz article on height information as such. I think you are reading far too much into his use of the expression "3D" in one sentence. Can you point me to any passage in which he refers to the encoding of height informaiton explictly?

Me justify that 3D includes height? Je refuse.

Some more references:

https://www.linkwitzlab.com/
"Hopefully these designs inspire a fresh look at what is possible in terms of believable 3D from two speakers before everyone launches off into 3D over headphones with their unnatural distance distortion."

https://www.linkwitzlab.com/Store/LXmini.htm
"The LXmini was designed to render sound recordings with clarity, neutrality and with detailed, 3D-like imaging, when properly set up."

https://www.linkwitzlab.com/LXmini/Introduction.htm
"With directivity controlled in this way the LXmini becomes much less sensitive to room placement, while also gaining in 3D imaging precision."

https://www.linkwitzlab.com/What_is_new_at_linkwitzlab.htm
"A prototype is already running and impresses me with its fast dynamics, 3D rendering and expected extension of bass"

https://www.linkwitzlab.com/Ping_tests/room_reflections.htm
"The data here for what is empirically a very good listening environment and setup for 3D phantom scene generation with the LX521 can serve as reference information ... This could become a major step to firmly establish the optimum loudspeaker and room combination for creating a 3D stereo illusion. But it could also turn out that the brain uses cues, which are not obvious from these data."

https://www.linkwitzlab.com/music.htm
"I find these transmissions very revealing of what the recording engineer or producer has decided to capture and transmit. In particular the real or unreal spatial context for a particular piece of music, the 3D aspects in the recording, are readily recognized even at 128 kbps."

Regards, D

 

[tuga]

Me justify that 3D includes height? Je refuse.

Some more references:

https://www.linkwitzlab.com/
"Hopefully these designs inspire a fresh look at what is possible in terms of believable 3D from two speakers before everyone launches off into 3D over headphones with their unnatural distance distortion."

https://www.linkwitzlab.com/Store/LXmini.htm
"The LXmini was designed to render sound recordings with clarity, neutrality and with detailed, 3D-like imaging, when properly set up."

https://www.linkwitzlab.com/LXmini/Introduction.htm
"With directivity controlled in this way the LXmini becomes much less sensitive to room placement, while also gaining in 3D imaging precision."

https://www.linkwitzlab.com/What_is_new_at_linkwitzlab.htm
"A prototype is already running and impresses me with its fast dynamics, 3D rendering and expected extension of bass"

https://www.linkwitzlab.com/Ping_tests/room_reflections.htm
"The data here for what is empirically a very good listening environment and setup for 3D phantom scene generation with the LX521 can serve as reference information ... This could become a major step to firmly establish the optimum loudspeaker and room combination for creating a 3D stereo illusion. But it could also turn out that the brain uses cues, which are not obvious from these data."

https://www.linkwitzlab.com/music.htm
"I find these transmissions very revealing of what the recording engineer or producer has decided to capture and transmit. In particular the real or unreal spatial context for a particular piece of music, the 3D aspects in the recording, are readily recognized even at 128 kbps."

Regards, D

It’s a figure of speech.

Figure of speech, any intentional deviation from literal statement or common usage that emphasizes, clarifies, or embellishes both written and spoken language.

 

[adamdea]

Me justify that 3D includes height? Je refuse.

Some more references:

https://www.linkwitzlab.com/
"Hopefully these designs inspire a fresh look at what is possible in terms of believable 3D from two speakers before everyone launches off into 3D over headphones with their unnatural distance distortion."

https://www.linkwitzlab.com/Store/LXmini.htm
"The LXmini was designed to render sound recordings with clarity, neutrality and with detailed, 3D-like imaging, when properly set up."

https://www.linkwitzlab.com/LXmini/Introduction.htm
"With directivity controlled in this way the LXmini becomes much less sensitive to room placement, while also gaining in 3D imaging precision."

https://www.linkwitzlab.com/What_is_new_at_linkwitzlab.htm
"A prototype is already running and impresses me with its fast dynamics, 3D rendering and expected extension of bass"

https://www.linkwitzlab.com/Ping_tests/room_reflections.htm
"The data here for what is empirically a very good listening environment and setup for 3D phantom scene generation with the LX521 can serve as reference information ... This could become a major step to firmly establish the optimum loudspeaker and room combination for creating a 3D stereo illusion. But it could also turn out that the brain uses cues, which are not obvious from these data."

https://www.linkwitzlab.com/music.htm
"I find these transmissions very revealing of what the recording engineer or producer has decided to capture and transmit. In particular the real or unreal spatial context for a particular piece of music, the 3D aspects in the recording, are readily recognized even at 128 kbps."

Regards, D

Still no actual mentions of height. I noticed that in the orginal article he dealt with other aspects (but not elevation) which is a bit of a give away.

Having thought about the "height information" question- the following point occured to me. If you know how wide an upright object of known shape is (say a piano), you have a pretty good idea how tall it is. On that basis you could say that in that sense height information could be conveyed implicitly. And if you did say that, I would agree.