Blue or red pill?

[mansr]

30 minutes ago, adamdea said:

Are you aware of the phenomenon of visual steering. I am not suggesting that blind people can't localise sounds- I'm saying that normally sighted individuals in everyday life do use vision to localise sounds, or rather their experience of sound localisation is  at least partially determined (where they can see the presumed source of the sound) by the visual information. I'm not suggestign that we can;t or don;t use audio information. But If the audio and visual information don't match the vision wins (interestingly the other way round for timing information). And that means you "hear" the sound in a different place from the real place it is coming from and do not perceive a struggle. In that sense you really do "hear" things you see.  And this is just an example of multisensory integration. Our "hearing impressions" come at least in part from non audio sources.

It can be useful to look at situations where the sensory integration breaks down. For example, if a jet aeroplane passes nearby, you can clearly hear the sound coming from well behind the plane, even though we know it to be emanating from the plane itself. Another example is when observing someone at a distance striking something with a hammer or slamming a door. We see the even take place, and somewhat later we hear the sound. In both instances, the discrepancy between the visual and auditory inputs is too large, and the brain gives up trying to reconcile them. On the flip side, if a plausible visual event matches a sound, the brain will readily associate the two, which is how film dialogue is plainly perceived as coming directly from the on-screen image of the actors even though in reality is is all supplied by a single centre speaker.

[mansr]

9 minutes ago, semente said:

It would be interesting to record with a vertically oriented ORTF setup and then play back with one speaker on the lower and the other on the top shelf of a bookcase.

I'm willing to bet that such a setup will give you height but no width - all instruments will be virtually playing from a vertical line positioned between the two speakers. It might be more difficult or near impossible to determine depth.

It might also be interesting to try my proposed three-mike ORTF played back with a raised centre speaker.

[adamdea]

16 minutes ago, mansr said:

It can be useful to look at situations where the sensory integration breaks down. For example, if a jet aeroplane passes nearby, you can clearly hear the sound coming from well behind the plane, even though we know it to be emanating from the plane itself. Another example is when observing someone at a distance striking something with a hammer or slamming a door. We see the even take place, and somewhat later we hear the sound. In both instances, the discrepancy between the visual and auditory inputs is too large, and the brain gives up trying to reconcile them. On the flip side, if a plausible visual event matches a sound, the brain will readily associate the two, which is how film dialogue is plainly perceived as coming directly from the on-screen image of the actors even though in reality is is all supplied by a single centre speaker.

Yes exactly.

The rubber hand experiment is fun too.

The interesting point for me is that we are always fitting information into a model (usually unconsciously)

I was reading a layman's book on neuroscience recently which iirc had some example where perceptual information was interpreted differently across cultures. I don't think it was pareidolia (although that is interesting enough)

[davide256]

The sad fact is the construction of most public venues muddies spatial perception for acoustic instrument location on stage. It takes an acoustically well designed hall to provide the clarity needed for knowing where on stage an instrument is playing vs just the general sense of stage left, center, right. The times I've found it easy to locate instruments, the halls were built before electronic amplification existed.

[Audiophile Neuroscience]

3 hours ago, mansr said:

Lateral positioning in stereophonic reproduction is an illusion, yes, Importantly, it is a deliberately crafted illusion. Depth and height, on the other hand, are unintentional illusions.

 

Good, Illusions all, agreed.At this stage lets just stay with depth.

 

Whether the illusion is intentional or not, crafted or accidental, added in the recording or after in post, is totally immaterial from a scientific point of view.They are real physical stimuli acting as real perceptual cues. Sticking your head in the sand and selectively ignoring some stimuli and not others will obviously lead to errors in any experimental setup.

 

 

Quote

The recording makes no attempt to capture the actual depth and height of sound sources. The cues it can convey do not in themselves actually indicate depth or height, not at a live event and not in a recording.They do, however, trigger the same perceptions in either situation.

 

Recording mics don't "attempt" anything, they passively transduce sound into electrical energy within their parameters of operation and how they have been setup.

 

As already said, the 2 channel recording does capture depth cues ie amplitude/volume, reverberation and frequency information.These are the physical cues subserving depth whether live or recorded. That is the evidence. Again, did you not look at the link provided by @esldude?

 

3 hours ago, mansr said:

They do, however, trigger the same perceptions in either situation.

 

Exactly - The physical cues conveyed trigger the same perceptions in either the live situation or recording. That would be physically impossible, objectively speaking, if the cues weren't there in the recording/signal in the first place. If there you should be able to measure them according to your previous statements. They are there, that is no longer the question....and they can be heard. The question is can you measure what is there.

[Audiophile Neuroscience]

2 hours ago, adamdea said:

Are you aware of the phenomenon of visual steering. I am not suggesting that blind people can't localise sounds- I'm saying that normally sighted individuals in everyday life do use vision to localise sounds, or rather their experience of sound localisation is  at least partially determined (where they can see the presumed source of the sound) by the visual information. I'm not suggestign that we can;t or don;t use audio information. But If the audio and visual information don't match the vision wins (interestingly the other way round for timing information). And that means you "hear" the sound in a different place from the real place it is coming from and do not perceive a struggle. In that sense you really do "hear" things you see.  And this is just an example of multisensory integration. Our "hearing impressions" come at least in part from non audio sources. [for another example IIRC different coloured objects sound louder or less loud according to their colour.]

 

None of this is revelatory. Yes of course there is multisensory integration together with affective, cognitive/evaluative and other integratory processes that contribute to perception. And yes of course illusions occur. That does not mean that iilusion is the main way normal people perceive information in any of the senses including hearing, they are distorted perceptions.Vision is not the main and normal way we perceive sound.Again, evidence please.

 

2 hours ago, adamdea said:

Do I need to provide evidence of a well understood process? I'm not trying to raise the temperature but I can't help wondering whether you have taken into account multisensory integration in your starting point ie your model of how and to what extent we can localise sounds in real life.

 

You are kidding right, I could write a book on multisensory integration in both unimpaired and pathophysiological conditions. My post immediately above remains true. If you disagree,fine but please provide evidence.

 

2 hours ago, adamdea said:

 

It would not be wise to assume that everything we experience in real life which we perceive as an auditory phenomenon could be replicated using only sound information.

 

You misunderstand. It is mans assertion that if its audible then it is measurable.

 

 

2 hours ago, adamdea said:

As for the rest of your long post- in your debate with Mansr you seem unwilling to consider the distinction between the question of whether spatial information can be encoded and the question of whether it can or might be perceived (sometimes) as result of the sensory system.

 

You misunderstand. I am saying no such thing.You are saying visual cues are the most important way we localize sound. I am saying that apart from some specific illusions (by definition distortions) you are wrong.

 

2 hours ago, adamdea said:

 

He is saying that you can't encode height information.

 

Again, I am not talking about height (yet).

 

2 hours ago, adamdea said:

As for depth I think we are agreed that by manipulating the amount of direct and indirect sound we can create some sort of sense of depth. But does that mean that it is because of reliably encoded auditory information that some people feel that they can tell in an orchestral recording that the flutes are in the middle 4 feet behind the middle of the strings just like they can when they see a live concert. 

 

 

Depth perception comes from a number of cues not just combinations of direct/indirect sound.2 channel encoded auditory information is inferior to 3 channel as has been demonstrated. Again, yes, depth information IS in 2 channel recordings as has been demonstrated with evidence.

 

I am sorry but if you play the science card you need to consider evidence not just compelling theories that suit your beliefs.

[Audiophile Neuroscience]

2 hours ago, mansr said:

It can be useful to look at situations where the sensory integration breaks down. For example, if a jet aeroplane passes nearby, you can clearly hear the sound coming from well behind the plane, even though we know it to be emanating from the plane itself. Another example is when observing someone at a distance striking something with a hammer or slamming a door. We see the even take place, and somewhat later we hear the sound. In both instances, the discrepancy between the visual and auditory inputs is too large, and the brain gives up trying to reconcile them. On the flip side, if a plausible visual event matches a sound, the brain will readily associate the two, which is how film dialogue is plainly perceived as coming directly from the on-screen image of the actors even though in reality is is all supplied by a single centre speaker.

 

Strange that you view this as a breakdown of sensory integration.Some might be valid perceptions and others illusions. I am not an expert on planes but it has been said that it is the turbulence in the air that creates that sound.There is also the doppler effect happening. Seeing before hearing distant noise producing events is no mystery and in what way do you mean the brain gives up reconciling them that defines a breakdown in sensory integration?

[mansr]

53 minutes ago, Audiophile Neuroscience said:

As already said, the 2 channel recording does capture depth cues ie amplitude/volume, reverberation and frequency information.These are the physical cues subserving depth whether live or recorded. That is the evidence.

How obtuse is it possible to be? beerandmusic is a little worse, but you're getting pretty close.

 

A stereo recording unambiguously captures the direction of a sound source. Depth is not captured whatsoever. When we perceive a reproduced sound as distance, it is because this fits our mental model of the world, not in any way because the recording actually contains depth information. Why do you persistently refuse either understand or accept this distinction?

[adamdea]

53 minutes ago, Audiophile Neuroscience said:

 

Depth perception comes from a number of cues not just combinations of direct/indirect sound.2 channel encoded auditory information is inferior to 3 channel as has been demonstrated. Again, yes, depth information IS in 2 channel recordings as has been demonstrated with evidence.

 

I am sorry but if you play the science card you need to consider evidence not just compelling theories that suit your beliefs.

I'm not going to waste time going over digressions. This bit is the key to the impasse you are having with mansr and there is an easy way out of it if you think about it.

Consider the example of the (potential) relative depth cue by relative amplitude. Mansr's point is that because this information is wholly ambiguous (the two sounds sources might be at the same distance or..) you can't measure using only the data and different people will interpret that data differently for reasons which you apparently understand. So you can't measure the depth cues as such. It's  a category error. What would the units of this measurement be. 

That isn't the same as saying that once you had identified what property caused this perception in this person you could then in principle measure that property. But it would have to be specific to the application of this recording to this person. There isn't a special code for 2ft back and slightly to the left.

[esldude]

Pardon me if I missed some comments.  Getting back to my supposition that vertically displaced microphones could create comb filtering that might mimic or display a sense of height though likely an inaccurate one. 

 

I crossed a pair of pencil condensers a few feet from a speaker.  Played white noise with everything gently rolled off below 4 khz.  First straight ahead then a vertical 30 degree angle, then 60 and then 90 degrees to the direction of sound.  I basically had a point source from the tweeter being recorded.  

 

There is some comb filtering in the combined result.  It is in or around the right area to possibly create height impressions.  There is none when there is no vertical offset to the mikes. 

 

OTOH, I don't hear any changes in height playing back those signals.  So any such thing might be too weak, or simply that it doesn't work that way in such a simplified manner.   So my supposition is at best a very weak maybe.  Not fully ruled out though right now appears not to be a factor in height sensed in some recordings.  

 

BTW, I used cardioid patterns.  The change in off axis FR is obvious.  The apparent position upon playback didn't change.  I repeated using a pair of omnis.  While you'll never record stereo like that it mostly eliminated off axis FR changes.  Still had comb filtering.  Still no apparent change in height with angle of offset. 

[semente]

15 minutes ago, esldude said:

Pardon me if I missed some comments.  Getting back to my supposition that vertically displaced microphones could create comb filtering that might mimic or display a sense of height though likely an inaccurate one. 

 

I crossed a pair of pencil condensers a few feet from a speaker.  Played white noise with everything gently rolled off below 4 khz.  First straight ahead then a vertical 30 degree angle, then 60 and then 90 degrees to the direction of sound.  I basically had a point source from the tweeter being recorded.  

 

There is some comb filtering in the combined result.  It is in or around the right area to possibly create height impressions.  There is none when there is no vertical offset to the mikes. 

 

OTOH, I don't hear any changes in height playing back those signals.  So any such thing might be too weak, or simply that it doesn't work that way in such a simplified manner.   So my supposition is at best a very weak maybe.  Not fully ruled out though right now appears not to be a factor in height sensed in some recordings.  

 

BTW, I used cardioid patterns.  The change in off axis FR is obvious.  The apparent position upon playback didn't change.  I repeated using a pair of omnis.  While you'll never record stereo like that it mostly eliminated off axis FR changes.  Still had comb filtering.  Still no apparent change in height with angle of offset. 

 

If my understanding is correct you can portray source vertical position and/or height if you place not just the mics but also one speaker above the other.

[esldude]

Just now, semente said:

 

If my understanding is correct you can portray source vertical position and/or height if you place not just the mics but also one speaker above the other.

I feel like I'm missing something in the question.

 

Do you mean using a vertical oriented extra mike fed to a different raised speaker?  It should work, but that isn't something I've done. 

[semente]

1 minute ago, esldude said:

I feel like I'm missing something in the question.

 

Do you mean using a vertical oriented extra mike fed to a different raised speaker?  It should work, but that isn't something I've done. 

 

I meant this:

 

3 hours ago, semente said:

It would be interesting to record with a vertically oriented ORTF setup and then play back with one speaker on the lower and the other on the top shelf of a bookcase.

I'm willing to bet that such a setup will give you height but no width - all instruments will be virtually playing from a vertical line positioned between the two speakers. It might be more difficult or near impossible to determine depth.

 

[esldude]

2 minutes ago, semente said:

 

I meant this:

 

 

Don't think that will work the way you are picturing it in your mind. 

 

There would be some height from one of the speakers having height.  It would not in my opinion, be like left to right stereo turned vertically.  

 

Think about this too.  A Calrec Soundfield microphone or its imitators consists of 4 microphones in a tetrahedron.  That can be mixed to create any type mike pattern aimed in any direction.  If you create and steer a virtual mike vertically you still don't record a kind of vertical stereo with those.  

 

I actually have intended to do some experiments with mixing in vertical oriented mics since back before Thanksgiving.  Other duties have kept me from it.  Eventually I will.  Some surround setups place cardioids pointed upward for the rear channels. 

[mansr]

13 minutes ago, esldude said:

Pardon me if I missed some comments.  Getting back to my supposition that vertically displaced microphones could create comb filtering that might mimic or display a sense of height though likely an inaccurate one. 

 

I crossed a pair of pencil condensers a few feet from a speaker.  Played white noise with everything gently rolled off below 4 khz.  First straight ahead then a vertical 30 degree angle, then 60 and then 90 degrees to the direction of sound.  I basically had a point source from the tweeter being recorded.  

That's fine for a test, but it's a pretty extreme setup. In a real recording situation, I'd be surprised to find any instruments more than 10° off horizontal.

 

13 minutes ago, esldude said:

There is some comb filtering in the combined result.  It is in or around the right area to possibly create height impressions.  There is none when there is no vertical offset to the mikes. 

What do you mean by combined result?

 

[semente]

9 minutes ago, esldude said:

Think about this too.  A Calrec Soundfield microphone or its imitators consists of 4 microphones in a tetrahedron.  That can be mixed to create any type mike pattern aimed in any direction.  If you create and steer a virtual mike vertically you still don't record a kind of vertical stereo with those.  

 

But did you try to reproduce the recording with the speakers aligned vertically and not horizontally?

 

I would like to know whether one would be able to get some sort of vertical position reproduction by using upper-left lower-left and upper-right lower-right channels/speakers.

[mansr]

9 minutes ago, esldude said:

I actually have intended to do some experiments with mixing in vertical oriented mics since back before Thanksgiving.  Other duties have kept me from it.  Eventually I will.  Some surround setups place cardioids pointed upward for the rear channels.

That's probably done in order to pick up predominantly reflected sounds which when played back on the rear speakers can create an impression of atmosphere or whatever you want to call it. Nothing to do with vertical positioning.

[mansr]

1 minute ago, semente said:

But did you try to reproduce the recording with the speakers aligned vertically and not horizontally?

 

I would like to know whether one would be able to get some sort of vertical position reproduction by using upper-left lower-left and upper-right lower-right channels/speakers.

Are you trying to reinvent Dolby Atmos?

[semente]

Just now, mansr said:

Are you trying to reinvent Dolby Atmos?

 

Not really.

Perhaps trying to conclude that one cannot get height with 2-channel stereo by showing that we would need 2 more channels.

 

Not that I really "need" to perceive height from my recordings. I am quite happy with what I have.

[rando]

Is this thread now about the audio equivalent of picture in picture?   Sorta like having ice dancing on full screen for the slips while really watching Super GS highlights.  

[esldude]

15 minutes ago, mansr said:

That's fine for a test, but it's a pretty extreme setup. In a real recording situation, I'd be surprised to find any instruments more than 10° off horizontal.

Yes, most of the time this is likely true.  Maybe not for sound reflected off of a church ceiling. 

15 minutes ago, mansr said:

 

What do you mean by combined result?

 

Well combining both channels.  The presumption was you hear both channels upon playback at each ear.  Combining the left and right channel you see some comb filtering that each individual channel does not have. 

[esldude]

13 minutes ago, mansr said:

That's probably done in order to pick up predominantly reflected sounds which when played back on the rear speakers can create an impression of atmosphere or whatever you want to call it. Nothing to do with vertical positioning.

Yes in the surround case it is.  I was planning on different experiments. 

 

J_J Johnstons perceptual sound field method of recording used an upward and downward facing microphone in the middle of a 5 mike array.  He has a formula for mixing in the vertical mikes into the 5 horizontal channels.  Apparently done correctly it at least adds space or volume to the perceived sound.  So I was going to experiment along those lines adding vertical mikes and just see if anything interesting results. 

[acg]

7 hours ago, mansr said:

If you measure the angle occupied by the moon when near the horizon and again when high in the sky, you get almost exactly the same value. Nevertheless, the rising or setting moon is often perceived as twice the size of the moon when (more or less) overhead. Atmospheric refraction can't explain this.

 

Atmospheric refraction perfectly explains this.  And I have measured it, more than once during my university degree, of both the sun and the moon.  They both move quite quickly and are relatively difficult to manually track through an instrument.

[adamdea]

2 hours ago, Audiophile Neuroscience said:

 

None of this is revelatory. Yes of course there is multisensory integration together with affective, cognitive/evaluative and other integratory processes that contribute to perception. And yes of course illusions occur. That does not mean that iilusion is the main way normal people perceive information in any of the senses including hearing, they are distorted perceptions.Vision is not the main and normal way we perceive sound.Again, evidence please.

 

 

I am sorry but if you play the science card you need to consider evidence not just compelling theories that suit your beliefs.

I did say that I wasn't going to pursue digressions but as long as it is understood that I have conceded that I was somehwat over -egging the pudding by saying that vision was the main source of sound localisation and only really saying that it is part of our everyday localisation (and that it tends to dominate actual auditory information in the case of conflict) then the main sources are listed here https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2674475/.

"Although sound sources can obviously be localized on the basis of auditory cues alone, localization accuracy improves if the target is also visible to the subject (Shelton & Searle 1980; Stein et al. 1989). This is an example of a more general phenomenon by which the central nervous system can combine inputs across the senses to enhance the detection, localization and discrimination of stimuli and speed up reactions to them

...Conflicting visual cues can also influence the perceived location of a sound source. Thus, the presentation of synchronous but spatially disparate visual and auditory targets tends to result in mislocalization of the auditory stimulus, which is perceived to originate from near the location of the visual stimulus (Bertelson & Radeau 1981). 

....Nevertheless, visual localization is normally more accurate than sound localization and therefore tends to dominate conflicts between the two modalities, thereby enabling, at least within certain limits, spatially misaligned cues to be perceived as if they originate from the same object or event. Repeated presentation of consistently misaligned cues results in a shift in the perception of auditory space that can last for tens of minutes once the visual stimulus is removed (figure 2). This is known as the ventriloquism after-effect and has been observed in both human (Radeau & Bertelson 1974; Recanzone 1998; Lewald 2002) and non-human (Woods & Recanzone 2004) primates. 

Again presumably you know all of this. It at the very least supports the conclusion that the everyday functioning of a normal person's experience of sound location is going to involve their vision (and quite a lot else besides).  

 

Even ignoring visual input to "hearing" It still seems to me to be obvious that the absence of the spectral cues arising from the hrtf (let alone the ability to cross check by tilting and moving ones head) severely limits the ability to encode and decode sounds to and from ordinary stereo recording as well as explaining how our perceptual system may make certain guesses based on spectral information. 

[mansr]

32 minutes ago, esldude said:

Well combining both channels.  The presumption was you hear both channels upon playback at each ear.  Combining the left and right channel you see some comb filtering that each individual channel does not have.

OK, I see what you're getting at. How does this compare to all the other comb filtering that is anyway present, in both the recording and playback venues? Did you take into account that the ears are not equidistant from the speakers and that they do not receive the same intensity from both?