What exactly is "imaging" ?

[Sue Pertwee-Tyr]

If you look at how speakers and speaker drivers produce sound and think how a phantom image is generated between two speakers (or on one of them) you'll understand why.

Maybe it’s me but no, sorry, not getting it. If they can create a credible phantom image between the speakers, why cannot this image also have a height aspect?

 

[herb]

My favourite seats in the Usher Hall are high above the violins, the sound is simply better up there. In the rear stalls the sound is amazingly similar to my kit.

 

[adamdea]

Yes, the diagram is clearly too simplistic. The reality is that we only have two ears and can't hear 'distance' in a simple way *even when listening at a live performance*. Instead, our brain infers it from the direct/reverb/(apparent)direction characteristics and having become familiarised with room acoustic effects.

Thus the usefulness of speakers and listening room with a high direct/reflected delivery. This reduces the 'confusing overlay' of the listening room's effect and allows you to hear the direct/reflected patterns of the recording or broadcast.

The skill here is in large part that of the producer who laid out and balances the mics as suitable for a venue. Or in some cases how that 'fake' an acoustic to give the impression of depth, etc.

I agree that there is no auditory equivalent of stereoscopic vision in terms of distance perception. Our real life distance perception is influenced IIRC but an awful lot of things. I spent some time reading the literature on distance/direction perception a while ago, just because. My recollection is that alot of the time we use other information including vision to produce our auditory mapping of sounds to place. Obviously the cinema centre channel/speech effect is well known. Other examples include the shouting/ whispering thing -you hear the former as being distant and the latter close if at the same amplitude and IIRC a bird song effect (you tend to hear it as being " up there"). So as everyone knows* the fact that one perceives a sound as coming from a particular location may have little to do wiht the sound waves reaching our ears.

Aside from that point. A stereo can create an illusion of objects in front of you on a left right plane for very specific reasons. However aside from this very specfic trick, a stereo cannot produce the auditory 3D sound cues (whether distinct objects or ambience) available in real life because it doesn't get the sound to come from the right direction. So all reflections in the recorded space come back and are recorded the same irrepective of direcition whereas as we all* know much of our ability to localise sounds (so far as it is actually based on auditory information- see above) comes from the effects ones ears and head have on sounds coming from different directions.

It does not necessarily follow thereofore that allowing the stereo recording to be "perfectly" reproduced by the speaker without those pesky room reflecions will actually make things better. They still won't be accurate and may actually be more confusing because you will then get no sound coming from behind you or from the side (which is actually going to be weird if you are in a room. ) In fact I had always taken it as a given that people generally don't like the sound of stereo speakers in anechoic chambers [reference anyone? I can;t remmber it]. It seems as though many/most people's ears prefer to have some bogus extra refections to having fewer (even if more acruate). I suspect that this is why recordings in the Albert hall sound surprisingly good compared with dry acoustics like the barbican, even though no one in their right mind would say that about the live sound.


*ie a handful of people on an audio forum

 

[tuga]

Maybe it’s me but no, sorry, not getting it. If they can create a credible phantom image between the speakers, why cannot this image also have a height aspect?

Every phantom image of identical spectrum will have the same height. With concentrical drivers this will be even more true.
It's differences in level between the two speakers that position the phantom image horizontally more to one or the other side. To produce both vertical and horizontal information you'd need 2 up and 2 down channels.

Two speakers only provide location on the X/horizontal plane if they're side by side or the Y/vertical plane if one is above the other.

 

[adamdea]

Maybe it’s me but no, sorry, not getting it. If they can create a credible phantom image between the speakers, why cannot this image also have a height aspect?

The answer is the same as it was the last 250 times this came up.

 

[pickwickpapers]

There is also artistic license with entirely studio-created music like rock/pop, things like Dark Side Of The Moon through to anything by Aphex Twin, Radiohead etc just don’t exist in the real world and are assembled in the studio overdub by overdub. To my mind it is unforgivable with say a live Keith Jarrett recording, I really don’t want a wide stereo piano and rock-style kick & snare centre, hi-hat right, ride left, toms panned across the stage there as frankly it sounds bloody daft.

It makes me chuckle when any one says something like ‘listen to the focus on her voice’ with something like a Norah Jones album. Any decent system should be able to sort out anything up to four or five distinct vocal layers overdubbed and slightly panned to give the sound she wanted. I have no issue with it in this sort of again studio-created music, but jazz and classical should in most cases be a recording of an actual event. I guess things blur there too, e.g. Miles and others fusion output (which I love), Stockhausen, John Cage etc.

Close-mic’d rock drum kits and wide stereo pianos do definitely trigger me in non-rock forms. I far prefer the ‘60s Blue Note etc approach of chucking a couple of mics in front of the kit and plonking it at one side of the soundstage in a coherent acoustic space. Just sounds way better to my ears. So much more real and believable.

I agree - but have to say, that as something of a drummist once upon a time, recordings never sound 'right' to me without the kit pretty much in the centre.
Stuff like the CTI stereo recordings, for example, with drums often shunted right over, bug me, I have to admit.

However, in terms of studio recordings of drumkits, one thing that never ceases to amaze me, in terms of projecting anything approaching a 'believable' stereo image, is the strong tendency for kits to be mixed as tho' from the drummer's position behind the kit.

Most drummers are right-handed, and even today, most kits are set up with the hi-hat in a conventional position on their left hand-side.
Yet the majority of recordings have the hi-hat on the left-hand side of the stereo image - and the rest of the kit following suit. At least of the recordings i've listened to (of 70s vintage and later).

 

[tuga]

I agree that there is no auditory equivalent of stereoscopic vision in terms of distance perception. Our real life distance perception is influenced IIRC but an awful lot of things. I spent some time reading the literature on distance/direction perception a while ago, just because. My recollection is that alot of the time we use other information including vision to produce our auditory mapping of sounds to place. Obviously the cinema centre channel/speech effect is well known. Other examples include the shouting/ whispering thing -you hear the former as being distant and the latter close if at the same amplitude and IIRC a bird song effect (you tend to hear it as being " up there"). So as everyone knows* the fact that one perceives a sound as some from a particular location may have little to do wiht the sound waves reaching our ears.

Aside from that point. A stero can create and illusion of objects in front of you on a left right plane for very specific reasons. However aside from this very specfic trick, a stereo cannot produce the auditory 3D sound cues (whether distinct objects or ambience) available in real life because it doesn;t get the sound to come from the right direction. So all reflections in the recorded space come back and are recorded the same irrepective of direcition whereas as we all* know much of our ability to localise sounds (so far as it is actually based on auditory information- see above) comes from the effects ones ears and head have on sounds coming from different directions.

It does not necessarily follow thereofore that allowing the stereo recording to be "perfectly" reproduced by the speaker without those pesky room reflecions will actually make things better. They still won't be accurate and may actually be more confusing because you will then get no sound coming from behind you or from the side (which is actually going to be weird if you are in a room. ) In fact I had always taken it as a given that people generally don't like the sound of stereo speakers in anechoic chambers [reference anyone? I can;t remmber it]. It seems as though many/most people's ears prefer to have some bogus extra refections to having fewer (even if more acruate). I suspect that this is why recordings in the Albert hall sound surprisingly good compared with dry acoustics like the barbican, even though no one in their right mind would say that about the live sound.


*ie a handful of people on an audio forum

As you say, 2-channel (or multi) stereo over speakers is incapable of recreating the original sound field because direct, reflected and reverberation all come from the same two sources located in front of the listener.
The most obvious example is audience noise and clapping which comes from the space between the two speakers instead of from the sides and from behindthe listener.

The only way to recreate the original soundfield would be to sit inside a spherical array of speakers with as many channels as those captured by a spherical mic array:

 

[tuga]

Those with small standmounts could try placing them vertically aligned some 5 or 6 ft apart and give it a go, though the research paper I posted earlier claims that our ability to discern location differences in the vertical plane is much weaker.

 

[Big Tabs]

Anecdotal fantasy... Or rubbish, if that's your thang.

I am not sure, but I think Oxymoron.

 

[tuga]

I am not sure, but I think Oxymoron.

“telling it like it isn’t...”

 

[adamdea]

Those with small standmounts could try placing them vertically aligned some 5 or 6 ft apart and give it a go, though the research paper I posted earlier claims that our ability to discern location differences in the vertical plane is much less accurate.

That paper can be found here incidentally.
https://core.ac.uk/download/pdf/145239364.pdf
leaving aside the lack of accuracy/sensitivity , maybe the key take home from the paper for this thread is the section on how you perceive height and front back/ direction and how you resolve the ambiguities in real life. Once that is understood then the problem of how to convey the relevant information to a listener from a recording should become clear. That and understanding how stereo actually works, and you're home.

 

[adamdea]

The only way to recreate the original soundfield would be to sit inside a spherical array of speakers with as many channels as those captured by a spherical mic array:

or possibly with headphones, a binaural recording made either with a dummy head which eaxctly matched yours or an ordinary binaural recording processed using your precise HRTF. On top of that I guess you ideally need head tracking processing .

 

[darrenyeats]

I am struggling to grasp what you're saying tuga.

It's like saying you can't hear height in a binaural recording because there is no sound source above. It's patent nonsense.

No adam, you don't need your exact HRTF. Close enough will do.

What is hard to understand about this?

Again, no-one is claiming the mechanism is technically accurate - just that there is a mechanism (or mechanisms) causing an illusion.

I hear this all the time. More so with my eyes closed BTW. If you're saying I won't hear it free field or in an anechoic chamber, then maybe - not sure.

 

[adamdea]

I am struggling to grasp what you're saying tuga.

It's like saying you can't hear height in a binaural recording because there is no sound source above. It's patent nonsense.

No adam, you don't need your exact HRTF. Close enough will do.

What is hard to understand about this?

Again, no-one is claiming the mechanism is technically accurate - just that there is a mechanism that produces an illusion.

Darren you are being obtuse. You can only apply an HRTF using a dummy head in a recoridng or by adding HRTF processing. You can't get it in an ordinary stereo recording.
On top of that even if you did do it it wouldnt work when reproduced with speakers only with headphones, because with speakers the sound will still be coming from the front so that "coming from the top" dummy head spectral cues will be overlayed on top by "coming from the front" real spectral cues.

 

[darrenyeats]

Binaural/HRTF would be a technically accurate illusion.

Technically inaccurate illusions are still possible.

 

[adamdea]

WE need to disinguish between ordinary stereo speaker systems on the one hand and binaural recordings with headphones on the other. In the case of the former there can't really be encoded height information although you can of course perceive all kinds of weird effects throuhg brain processing like hearing birdsong coming from high up.

And for the former you really do need (at least for most people) a pretty accruate personal HRTF. I've tried loads of binaural recordings and I rarely get any "in front" sense

 

[davidsrsb]

That paper can be found here incidentally.
https://core.ac.uk/download/pdf/145239364.pdf

Thanks for that link.
So I am correct that I can estimate elevation from spectral cues and then get better accuracy with dynamic cues.
My thoughts are that subtle frequency response changes can affect sensed height and the effect will vary across the panning
Nodding your head will get misleading dynamic cues depending on reflected sound.
As this has nothing to do with what was recorded on a studio recording, it will be a lottery.

 

[Big Tabs]

“telling it like it isn’t...”

You noticed?

Like, ‘Here’s a true story that I just made up.’

Simple things.

Anyhoo, whether anecdotal twaddle or other - I perceive width, height and depth when listening to music. Always have done.
I thought everyone did the same, but apparently not.

There is no graph or paper to prove or disprove anything, it is just what I hear.

 

[tuga]

You noticed?

Like, ‘Here’s a true story that I just made up.’

Simple things.

Anyhoo, whether anecdotal twaddle or other - I perceive width, height and depth when listening to music. Always have done.
I thought everyone did the same, but apparently not.

There is no graph or paper to prove or disprove anything, it is just what I hear.

I believe that we are talking about different things and thus not really disagreeing.

There is no phantom image height, width or position in the vertical plane being reproduced by the stereo pair of speakers.
But you perceive these things.
What I wrote is that this is your imagination at work. It's a good imagination because it makes the listening experience more real, in my opinion.



P.S.: I expect the kick drum to be on the floor so that's where I perceive it to be positioned, whilst vocals, a solo violin or cybals I tend to perceive above the top of the speakers.

 

[darrenyeats]

A couple of years ago I attended a demo and with one song I heard a high hat 1.5m above and 1m to the left of the left speaker. Thinking it was comb filtering or some other weird effect I deliberately shook my head and repositioned myself. No, still there in exactly the same spot.

This was a precisely located example of something that happens all the time IME.

That is an illusion, not imagination. Now if you're saying that an accurate illusion would have been the high hat in some other place, then I am willing to accept that, since the combination of recording acoustic, mic placement, speaker placing and listening room acoustic is likely to have unknown cumulative effect! Maybe another situation would have no effect, or a different effect, with the same recording. But if you say no height illusion can be conveyed by a stereo recording using speakers, you are wrong.