Blue or red pill?

[STC]

3 minutes ago, gmgraves said:

No, it's not a "defect of stereo". It's a defect of multi-miked, multi--channel monaural sound. Stereo uses TWO microphones, one for each channel and there is no bleed. In stereo, each mike picks up the entire ensemble, just from a different perspective. The term has, over the years been perverted to mean two playback channels, but that has little to do with stereo.

 

You can create stereo recordings even with one microphone by mixing different instrument and pan potting  them. 

[gmgraves]

2 minutes ago, STC said:

 

You can create stereo recordings even with one microphone by mixing different instrument and pan potting  them. 

Godammit, STC, That's NOT STEREO. That's multi-channel\multi-miked mono! Just calling it stereo doesn't make it so!

[STC]

25 minutes ago, gmgraves said:

Godammit, STC, That's NOT STEREO. That's multi-channel\multi-miked mono! Just calling it stereo doesn't make it so!

 

It doesn’t matter whether they were multi miked or recorded at different time. What matters is if you are good recording and mixing engineer, you could pan pot those recording and create a stereo recording which when played over a stereo system will still create the illusion of soundstage and phantom image. Do you think norah jones Come away vocal was recorded with a stereo microphone?

[Ralf11]

Let's go with stereo vs. stereophonic recording

[fas42]

1 hour ago, STC said:

 

It doesn’t matter whether they were multi miked or recorded at different time. What matters is if you are good recording and mixing engineer, you could pan pot those recording and create a stereo recording which when played over a stereo system will still create the illusion of soundstage and phantom image. Do you think norah jones Come away vocal was recorded with a stereo microphone?

 

Whether it is a purist stereo recording, or a confection, matters little in terms of hearing space, and depth in the recordings. If just a stereo pair microphone, single take, then everyone is in the same room - if otherwise, then there are or could be multiple spaces, of various sizes, positions, and character in front of one - all at the same time. This might sound confusing, to the ear, if the replay if not of a good standard - assuming such is in fact  the case, then it's easy to "grok" the individual spaces - if well done in the mastering then this is a very rich, satisfying world to explore - like investigating a well thought out garden, .

[beerandmusic]

9 hours ago, fas42 said:

 

 

 

Convincing means exactly that - that you are unable to determine whether the sound is from, say a live individual singing, or a recording.

I just have never heard or believe any system can ever be engineered that will convincingly sound like live...

[adamdea]

11 hours ago, esldude said:

About height info and comb filtering etc.  Probably upon reflection the chances spurious conditions accidentally show height in a recording are smaller than I previously believed.  The research where they put inserts into people's outer ears corrupted their height and directional perception for quite awhile.  They would have still had comb filtering, just different spectra of filtering going on.  Yet it ruined perception for a period of time. 

 

So the chances accidental comb filtering would make one hear height are probably down near the zero chance end of probabilities with recordings.  Again, the best evidence to the contrary would be simple recordings that had height perception with minimal processing.  Does anybody have some they can point to for us to listen to and decide?

Sorry if I'm being thick, but I can't follow your reasoning in the first paragraph. Can you remind me of the link you are referring to  I would expect that the gross effect of interfering with the subject's hrtf would temporarily override their ability  to judge height and depth precisely becasue the model of the world into which you put any input has now been disrupted (but not ILD and ITD to any great extent so left right should still be ok). This seems to me to be consistent with the ventroliquism after effect. The difference is that the ventroliquism effect is the result of inserting perceptually dominant visual information into the model whereas this involves jamming the purely auditory part of the model. I would expect it to settle down eventually once people had time to recalibrate.

 

If you were referring to this 

https://www.ncbi.nlm.nih.gov/pubmed/10196533

then the results are consistent with that. It seems to me that there is a big danger in trygin to analyse perceptual effects without taking into account the learning aspect. Our brain does not perceive auditory events at time t based only on information acquired at time t. This makes careful analysis of cause and effect tricky.

 

btw without wishing to flogh a dead horse it occurred to me that this article contains a pretty comprehensive summary of the position about hearing depth perception (not stereo)

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4744263/

[adamdea]

33 minutes ago, adamdea said:

btw without wishing to flogh a dead horse it occurred to me that this article contains a pretty comprehensive summary of the position about hearing depth perception (not stereo)

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4744263/

There's some very interesting stuff about the effect of distance estimation on what you think you know about the source 

Estimates of the distance to a talker increase systematically at a fixed location as the speakers’ vocal effort increases from conversational to shouted, whereas distance estimates for whispered voices decrease (Brungart & Scott, 2001; Gardner, 1969). Philbeck and Mershon (2002) confirmed that the source familiarity effects observed in these studies were due to long-term knowledge of the typical characteristics of speech rather than experimental context (comparison of stimuli across trials), by showing that shouted voices were reported as farthest and whispered voices closest even upon first stimulus presentation. Thus, source familiarity can provide absolute auditory distance information (Mershon & Bowers, 1979).

This explains why the whispered voice in "riders on the storm" seems to be whispering in your ear. It's becasue your brain "knows" how to interpet a whisper. If he was whispering from far away you wouldn;t be able to hear him would you? So it must be coming from really near. This illustrates the critical point that you interpret distance information using something other than the recording itself. That's why you can't "measure" the depth information in the recording- it's not in the recording it's in your head. 

[semente]

9 hours ago, STC said:

 

There is a reason why binaural recordings will not work well or more correctly to its true potential with loudspeakers. It is law of physics.  Chesky uses filter for their binaural recordings for loudspeaker playback.  Otherwise, it will sound just like ordinary stereo recordings.  In any case, the  true potential of the binaural recordings of Chesky cannot be heard with conventional setup.

 

“Because loudspeaker-crosstalk of conventional stereo interferes with binaural reproduction, either headphones are required, or crosstalk cancellation of signals intended for loudspeakers such as Ambiophonics is required.”

 

from Wiki. 

 

 

 

One could also try placing a pannel (or a matress?) vertically between the speakers.

 

 

http://haute.fidelite.com.online.fr/coin-integriste/arolio_systeme/index3.html

 

 

Tried it once with a large sofa seat cushion and it worked nice even with normal stereo (I only have a couple of binaural recordings), but you really need to close your eyes if you're to get any enjoyment from the listening session.

[STC]

3 minutes ago, semente said:

Tried it once with a large sofa seat cushion and it worked nice even with normal stereo (I only have a couple of binaural recordings), but you really need to close your eyes if you're to get any enjoyment from the listening session.

 

Really? It worked well? Physical isolation is not easy.  Care to explain you proper setup?  Thanks. 

[semente]

6 minutes ago, STC said:

 

Really? It worked well? Physical isolation is not easy.  Care to explain you proper setup?  Thanks. 

 

Look at the picture, it wasn't much different.

I used a pair of smallish standmounts, close together, axis parallel, in a near-field setup, away from boundaries.

This was long ago. Haven't tried it since.

[adamdea]

On 2/22/2018 at 12:18 AM, gmgraves said:

I think I did that already. HRTF doesn't come into play at the CAPTURE of the sound. How could it? there is no head, no ears there! The microphones pick-up various phase, timing, and intensity cues that when played back on speakers involve the listener's head The speakers reproduce (in as far as they can do so) the sound field captured by the microphones and the head and ears of the listener pick-up that sound field just like they would were they in the venue listening to the performance live. Seems pretty straightforward to me.

Sorry if it seems like people are trying to pick an argument, or disrespect your recording expertise, but I think I've worked out where it is that the disagreement or mutual stand off lies and I would just like to have one more go at seeing whether it can be resolved. I have no doubt that your stereo recordings are excellent and would love to hear them. But I really do think there is a chance that we can be made to see eye to eye here.. 

 

"The head and ears of the listener pick-up that sound field just like they would were they in the venue listening to the performance live".

 

You seem to be saying that

a. a pure stereo mic pair records everything direct and  indirect sound at that point in space.

b. if this is reproduced it will reach the listener and his head can do the work just as it would do if his ear was at the same pont in space as the mic.

 

But if we think of the sound at the point in space where the mic is, it is various sound pressure waves coming from all directions. It is mixed down to one single sound pressure by a conventional mic (not soundfield). Unfortunately the mic measurement, however accurate on its own terms is still just a mixdown: it does not contain all the information. That is it is not a measure of the sound pressure waves at that point. If it were the sound would not keep travelling past the mic.

 

An analogy would be a light meter which measured all the waveslengths of light at a single point coming from every direction, but mixed them all into one spectrum. That would in a sense be an accurate measurement of the ligth at that pojnt, but it would not help you to know what you would see if you were at that point. So the stero mics do not record the sound event at that point: they record a mixdown, something like a black and white photo, a partial recording. Crucially the sound reaching the mic from a point 5 m away and slightly up is recorded in just the same way as the info from 5 m away and exactly in front: they are all the same. And in any event those sounds are mixed up and you can;t take apart how the composite was formed.

 

As with a lot of this area, it makes you wonder how stereo works at all. The answer is that conveniently your ear also mixes down to one single sound pressure. at each time in your ears after rest of the the hrtf. to that extent it is similar. but the ear makes a mixdown which encodes some of the information about  distance height and direction- the sound from the back is different from the equivalent sound in the front because it goes through the back of your pinna not the front etc etc. Same for sound from high up. which in real life goes round the crown of your head and is filtered accordingly. That mixdown is done from the full multi-directional sound waves reaching the ear

 

The only part of that extra information which the mic picks up is the differences between channels (ILD and ITD) and  stuff that can be decoded from direct to reverb or volume differences etc. It is not the same mixdown as the mixdown which the ear would do at the same spot. You can;t tell front back direction from the recording alone. That is of course why playing ordinary stereo on ordinary headphones gives the sound in your head. 

 

We can now see that what the stereo mic records is neither the true soundfield at the point, nor is it the partial picture which the ear would take at that point. 

 

How can the missing information be recreated by the ear/hrtf when it is exposed to the sound coming from the speaker in front and on one plane only? That sound is now reaching the listener's head from  the direction of the speakers only (ignoring the listening room reflections for a moment).  It can use the ITD and IHDs to  reconstruct some left right information. The head cannot process the bit which originally came from 5m in front and slightly up differently from that which came from 5m away and on the level. It can only guess at the importance of the reverb in the recording becasue it can;t separate out the bits coming from the back wall in front, the side wall, the ceiling, the floor, the back wall behind the mic. Fortunately it doesn't get things back to front because the Hass effect locks the direction as being from the speakers. So fortunately as a result the whole thing can be made to broadly coincide with the real picture you would get in a recording where you sit in front of a orchestra and are interested in the sound in front of you (provided you sit in the sweet spot of course).

 

The stereo sound will therefore all come from the front and be roughly correct front  left/ right. But how do we go about decoding depth and height when the information in the original soundfield is not recorded and most of the the information your ears would have if you were sittign where the mics were is not recorded?

 

Now in this respect a purist stereo recording is good. The chances of making something of the information mess are improved from a bad multi mic affair because the very limited spatial information should at least be consistent. But what is recorded and what comes back to the listener is only vaguely like what their ears would have received if they were sittign at the same spot as the mic (and this is assuming a perfect stereo playback system and room).

 

That is the starting point for trying to work out what we experience when listening to the stereo recording and how that may or may not map to the real recorded space.

 

[STC]

41 minutes ago, semente said:

 

Look at the picture, it wasn't much different.

I used a pair of smallish standmounts, close together, axis parallel, in a near-field setup, away from boundaries.

This was long ago. Haven't tried it since.

 

Oh. You never mentioned this before and the physical barrier method was only way before the RACE was finally made available by Ralph.  That must be a very long ago.  

 

I tried with mattress but still couldn’t  get good isolation. Although I saw the potential but due to the mattress absorption of HF and sound was leaking to the other ear, it wasn’t too successful. That was around 2001. 

 

You must have performed it better than me.  I never reached that stage at that time.  Physically, despite my best effort I failed that I never seriously attempted again till 2009.

 

 

[adamdea]

1 hour ago, adamdea said:

There's some very interesting stuff about the effect of distance estimation on what you think you know about the source 

 

I just realised this is a typo it should have read

There's some very interesting stuff about the effect on distance estimation of what you think you know about the source 

[STC]

16 minutes ago, adamdea said:

I just realised this is a typo it should have read

There's some very interesting stuff about the effect on distance estimation of what you think you know about the source 

 

When it comes to stereo playback most of us imagine a lot of things about placement. In real life, some performed so poorly that they just couldn’t localize actual drop of coin in a dark room. 

 

Sorry no reference for that.  Just an experiment with fellow audiophiles after I got tired of some of them describing non existing placement with stereo playback.  

[esldude]

2 hours ago, adamdea said:

Sorry if I'm being thick, but I can't follow your reasoning in the first paragraph. Can you remind me of the link you are referring to  I would expect that the gross effect of interfering with the subject's hrtf would temporarily override their ability  to judge height and depth precisely becasue the model of the world into which you put any input has now been disrupted (but not ILD and ITD to any great extent so left right should still be ok). This seems to me to be consistent with the ventroliquism after effect. The difference is that the ventroliquism effect is the result of inserting perceptually dominant visual information into the model whereas this involves jamming the purely auditory part of the model. I would expect it to settle down eventually once people had time to recalibrate.

 

If you were referring to this 

https://www.ncbi.nlm.nih.gov/pubmed/10196533

then the results are consistent with that. It seems to me that there is a big danger in trygin to analyse perceptual effects without taking into account the learning aspect. Our brain does not perceive auditory events at time t based only on information acquired at time t. This makes careful analysis of cause and effect tricky.

 

btw without wishing to flogh a dead horse it occurred to me that this article contains a pretty comprehensive summary of the position about hearing depth perception (not stereo)

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4744263/

Your first link is there one I had in mind. If you read the full article upon disrupting the ear at first it took several weeks for height to be correctly heard. Now there was still comb filtering present it simply wasn't the pattern the brain was expecting.

 

So my reasoning is that some simple comb filtering in recordings due to microphone height difference would be too simple and too dissimilar to the complex filtering pattern from our outer ear to result in height perceptions.

[adamdea]

2 minutes ago, esldude said:

Your first link is there one I had in mind. If you read the full article upon disrupting the ear at first it took several weeks for height to be correctly heard. Now there was still comb filtering present it simply wasn't the pattern the brain was expecting.

 

So my reasoning is that some simple comb filtering in recordings due to microphone height difference would be too simple and too dissimilar to the complex filtering pattern from our outer ear to result in height perceptions.

I know that the article says that but it doesn't support your conclusion about comb filtering

The point is that the comb filtering might make a sufficient difference that when your brain applies it to its hrtf model it infers height. That has nothing to do with the long term effect caused by disrupting the model itself. In the latter cause all the bits of height information from all objects is disrupted  including the real ones in the real world, not just the odd one in a recording- that's why you entire system is confused!

 

In any event if you doubt this empirically I would take it up with JJ who will no doubt be able to tell you whether the comb filtering height effect it is experimentally verified 

[adamdea]

26 minutes ago, STC said:

 

When it comes to stereo playback most of us imagine a lot of things about placement. In real life, some performed so poorly that they just couldn’t localize actual drop of coin in a dark room. 

 

Sorry no reference for that.  Just an experiment with fellow audiophiles after I got tired of some of them describing non existing placement with stereo playback.  

Sure. No surprise. The interesting thing is that once one steps outside the whole unspoken and confused audiophile model of  hearing based on "common sense" extrapolation from what we know, then it all makes more sense. How often do most of us try and locate objects by sound alone in real life?

 

If you are not blind how do you generally estimate the position and location of objects? How accurately do you think you could estimate the distance and direction of an object you can see

a) by hearing alone

b) by sight alone?

 

If you can see an object then you will "hear" its location more accurately. 

 

If the information conflicts you will generally hear the object where you see it (ventriloquism effect)

 

Even if you can;t see it any more you may hear it where you saw it (ventriloquism after effect.)

 

Even if you think you know where the object is, you are likely to hear its location where you think you know it is (see examples about whispering voice)

 

I listen to a stereo recording and get a vague sense that the timpani are in the position they usually are in when I go to orchestral concerts. 

 

Yup.

[mansr]

24 minutes ago, adamdea said:

How often do most of us try and locate objects by sound alone in real life?

If I drop a small object on the floor, I generally try to listen for where it went so as to more easily find it. Quite often, it works.

[semente]

9 hours ago, STC said:

 

It doesn’t matter whether they were multi miked or recorded at different time. What matters is if you are good recording and mixing engineer, you could pan pot those recording and create a stereo recording which when played over a stereo system will still create the illusion of soundstage and phantom image. Do you think norah jones Come away vocal was recorded with a stereo microphone?

 

You can call perhaps call it a stereo mix. It's got the stereo effect but it's not a stereo recording.

[pkane2001]

48 minutes ago, adamdea said:

Sure. No surprise. The interesting thing is that once one steps outside the whole unspoken and confused audiophile model of  hearing based on "common sense" extrapolation from what we know, then it all makes more sense. How often do most of us try and locate objects by sound alone in real life?

 

If you are not blind how do you generally estimate the position and location of objects? How accurately do you think you could estimate the distance and direction of an object you can see

a) by hearing alone

b) by sight alone?

 

If you can see an object then you will "hear" its location more accurately. 

 

If the information conflicts you will generally hear the object where you see it (ventriloquism effect)

 

Even if you can;t see it any more you may hear it where you saw it (ventriloquism after effect.)

 

Even if you think you know where the object is, you are likely to hear its location where you think you know it is (see examples about whispering voice)

 

I listen to a stereo recording and get a vague sense that the timpani are in the position they usually are in when I go to orchestral concerts. 

 

Yup.

 

I've not been following the discussion about depth perception in recorded audio, as I generally agree there's usually very little 'real' depth information in most recordings, especially multi-mic, multi-track ones.

 

But, there is certainly something to the depth cues that might help locate a source of sound, while not precisely, but relative to other sources in the same recording. Call it reverb, echo, reflections, comb filters, combined with volume changes, these can all add to the sense of depth on a well made recording (or artificially manufactured depth on recordings where this is done through digital manipulation).

 

As I posted yesterday, I could clearly hear relative depth of instruments on a simple binaural recording when played back through loudspeakers. Some of the recordings I tried placed sound in front of my speakers, and this is the first time I've heard that effect. Again, these do not give a precise position, but relative to other sounds it makes for a nicely layered soundstage -- something we audiophiles love to hear.

 

Of course, I've also heard depth in other recordings, some seem to place sounds well beyond the speakers, others right at the speakers. An example of a manufactured sound stage depth I find amazing is on the Dire Straits On the Night live recording, f.i. The stage is placed so far behind my speakers, that it appears to come from the next room. As I understand it, that's just a post-processing trick, but makes for an impressive listening experience.

 

As far as measuring depth from a recording, I don't think that's possible with complex music, but it should be doable from some test signals, such as a recorded short pulse or a sweep.

[mansr]

4 minutes ago, semente said:

You can call perhaps call it a stereo mix. It's got the stereo effect but it's not a stereo recording.

We need to differentiate between stereophonic recording and stereophonic playback. One can have the latter without the former.

[semente]

1 minute ago, mansr said:

We need to differentiate between stereophonic recording and stereophonic playback. One can have the latter without the former.

Yes, that makes sense.

[esldude]

1 hour ago, adamdea said:

I know that the article says that but it doesn't support your conclusion about comb filtering

The point is that the comb filtering might make a sufficient difference that when your brain applies it to its hrtf model it infers height. That has nothing to do with the long term effect caused by disrupting the model itself. In the latter cause all the bits of height information from all objects is disrupted  including the real ones in the real world, not just the odd one in a recording- that's why you entire system is confused!

 

In any event if you doubt this empirically I would take it up with JJ who will no doubt be able to tell you whether the comb filtering height effect it is experimentally verified 

I don't think you understand what I'm saying. Posting from a phone.  I'll be more clear and deliberate in a later reply.

[semente]

20 minutes ago, pkane2001 said:

 

I've not been following the discussion about depth perception in recorded audio, as I generally agree there's usually very little 'real' depth information in most recordings, especially multi-mic, multi-track ones.

 

But, there is certainly something to the depth cues that might help locate a source of sound, while not precisely, but relative to other sources in the same recording. Call it reverb, echo, reflections, comb filters, combined with volume changes, these can all add to the sense of depth on a well made recording (or artificially manufactured depth on recordings where this is done through digital manipulation).

 

As I posted yesterday, I could clearly hear relative depth of instruments on a simple binaural recording when played back through loudspeakers. Some of the recordings I tried placed sound in front of my speakers, and this is the first time I've heard that effect. Again, these do not give a precise position, but relative to other sounds it makes for a nicely layered soundstage -- something we audiophiles love to hear.

 

Of course, I've also heard depth in other recordings, some seem to place sounds well beyond the speakers, others right at the speakers. An example of a manufactured sound stage depth I find amazing is on the Dire Straits On the Night live recording, f.i. The stage is placed so far behind my speakers, that it appears to come from the next room. As I understand it, that's just a post-processing trick, but makes for an impressive listening experience.

 

As far as measuring depth from a recording, I don't think that's possible from a music recording, but it should be doable from some test signals, such as a recorded short pulse or a sweep.

 

These are some of the probable mechanisms that I think allow us to perceive depth:

- we recognise certain ambience cues from our experience with live sound/reality (balance between direct and reflected sound)

- slight high frequency roll-off for distant sources

- instruments and vocals sound softer than others