Genelec on audio science

[Cosmik]

After all, stereo is the first, minimalist, step above mono

I'm still not convinced that there isn't something special about stereo. My view of stereo is not that it is merely twice as good as mono, but more akin to the rendering of a hologram - in effect replacing a thousand 'channels' if you want to see it that way. Those 'channels' are all in front of the listener, however (not including BACCH and related ideas).

It would appear obvious that if stereo sounds so good in one 'quadrant' then with four speakers around the listener it should be possible to create an all-round 'hologram', but I presume it doesn't work that way! What works with two speakers ahead of the listener doesn't work front-to-back, and nor can the placement be unambiguous if a single impulse is rendered from more than two speakers.

I presume this is why multi-channel then goes for many channels and speakers around the listener - and then it looks to someone like me as though a science fiction 'sound field renderer' has been replaced by something that is much more of a gimmick. Maybe a very enjoyable gimmick, but nonetheless not as sophisticated or miraculous as stereo.

 

[Theo]

Maybe a very enjoyable gimmick, but nonetheless not as sophisticated or miraculous as stereo.

Isn't stereo miraculous because it fits the number of ears we have?

I would say, stereo is an optimum. Mch may sound better, but is the step up in SQ worth the effort? How many channels are required to reconstruct a perfect sound field?
That is provided the recording is "perfect". It's been said above that it is a lot easier to make a good stereo image than a multichannel one.

 

[sergeauckland]

Around 2005, the German company Lawo demonstrated Sound Field Synthesis at one or two shows, together with the Fraunhofer Institute. This involved hundreds of channels of DSP, each channel feeding one small loudspeaker, with lower frequencies divided over several channels. The demonstration involved street noises including traffic with lots of LF, conversations as well as music, and used the Soundfield microphone as a source.

Apart from the sheer impracticality of several hundred loudspeakers, (and cost) it did allow genuine 'surround' over a wide area, but only in a large room and for a listener more or less central. When one got closer to the walls of the room, and thus nearer to the loudspeakers, it became far less convincing. I'm not sure it worked even as a 'proof of concept', and haven't heard any more about work in this area, but then I retired shortly after.
S

 

[Theo]

I would have guessed "infinite". Which your experience seems to confirm. The accuracy of soundfield reconstruction is an assymptotic function of the number of speakers... The 2.x (stereo with subwoofers), 5.1, 7.1, x.1 systems are an approximation, which may work under specific conditions (like a fixed listener position).

 

[Wombat]

Performance/environment>reality gap>recording> reality gap> reproduction(equipment, environment)> reality gap> the listener's hearing>reality gap.> listener perception> reality gap> other's perceptions.

Try empirically matching the beginning and the end of the chain. Show me.

 

[Blumlein 88]

I'm still not convinced that there isn't something special about stereo. My view of stereo is not that it is merely twice as good as mono, but more akin to the rendering of a hologram - in effect replacing a thousand 'channels' if you want to see it that way. Those 'channels' are all in front of the listener, however (not including BACCH and related ideas).

It would appear obvious that if stereo sounds so good in one 'quadrant' then with four speakers around the listener it should be possible to create an all-round 'hologram', but I presume it doesn't work that way! What works with two speakers ahead of the listener doesn't work front-to-back, and nor can the placement be unambiguous if a single impulse is rendered from more than two speakers.

I presume this is why multi-channel then goes for many channels and speakers around the listener - and then it looks to someone like me as though a science fiction 'sound field renderer' has been replaced by something that is much more of a gimmick. Maybe a very enjoyable gimmick, but nonetheless not as sophisticated or miraculous as stereo.

We have a zone of confusion of about 45 degrees directly right and left of our head. Sound in these areas is very difficult for us to place with any precision. In real life if something of interest happens in those zones we of course rotate our head to find out where it is.

One can record 4 quadrants of stereo. One could then hear in stereo any pair of speakers you face. However sitting face forward means those side signals aren't well located and rear signals are less well located. It is my understanding 4 closely spaced cardioids at 90 degree angles are often used for the basic background ambient sound. With other sources mixed into that. The technique is common in live or sporting events done with surround sound. Typically referred to as the IRT cross. It is set up like double ORTF with wider spacing. Double MS is a coincident method which decodes to 4 super-cardioids. Either of these can be used for basic ambiance with two or three other mics up front for the focused sound up front. This can work nicely. Also allows you to alter the ratio of ambient hall sound and the musicians you wish to focus on.

 

[Fitzcaraldo215]

Oh really ? But I have read also that the defects of a given loudspeaker would be much more audible if used alone in a mono configuration, than using the pair. So, if using two loundspeakers masks their defects, then a mono program listened to with two speakers should sound better, shouldn't it ?

I've just tried : it sounds good on the left speaker if I move my listening position in front of it. It also sounds good on the right one if I move the listening position towards the other side.
With both speakers, is sounds good too.

Since my room correction is the same on both channels, I can't comment on the general frequency response : it is necessarily flawed when I have only one speaker active. One of them is related with two low frequency room modes (55 Hz and 70 Hz), the other has only one (55 Hz). And my equalization corrects both of them on both channels at once.

The difference is that with both speakers, the sound is "wider". It is not concentrated inside a central phantom image. I can hear some high frequencies coming from the sides, although my configuration is narrower than the usual equilateral triangle.

So I guess that a single speaker is better if you need to create a central image, like with dialogs in movies, but two speakers are better to listen to music recorded in mono.
By the way, I've heard that using several subwoofers was a good idea, although they all play the same mono signal...

You should listen in the way you prefer. But, that extra "wider" image with high frequencies on the sides is clearly not on the mono recording, and whether or not it in any way replicates what was originally heard in the venue is pure guesswork. I venture it does not sound much like the original sound in the venue in any case. Those qualities you prefer are added to the mono sound on the recording by your speakers and room reflections, meaning they are effects or colorations. But, enjoy them, if you like them.

I believe the comparisons to which you refer about multiple speakers were about listening to mono sources with one speaker, to stereo sources with two, and to Mch sources with many speakers. I do not think you can generalize from that about mono heard from increasing numbers of speakers being increasingly preferable, certainly not in terms of reproducing what is on the recording with fidelity.

But, yes, I agree, stereo sound is preferable to mono, and Mch is preferable to stereo. And, yes, I believe we become less picky about the loudspeaker quality as the number of channels is increased, as Toole shows.

 

[svart-hvitt]

Very intersting thread, with a lot of information to process. Thank you all!
A comment about the thesis (unfortunately, couldn't read it as it is in german...): it would appear that according to the spectrum curves figures that the "best" recordings (at least below 3-4kHz) would be from 1976-1978 or 1985-1987... Any particular reason?
What is missing though, IMHO, is an analysis based on music style. I would bet that these curves would look different. The share of acoustic (jazz/classical...) or electronic (D&B for example) - which started to become a significant share of the music market in the 80's, may explain some of the general trend.
I am not a professional sound engineer, which would require some actual training, so this question maybe out of scope: isn't part of the mastering process to listen to the final mix through different types of equipment, making sure it sounds good on most, including small systems (i.e. lacking bass rendering), by far the most common in the general public? Would that explain the bass boost? What happened in 1976-1978 (and 1985-1987?)?

@Theo , you remarked that what "is missing though, IMHO, is an analysis based on music style. I would bet that these curves would look different. The share of acoustic (jazz/classical...) or electronic (D&B for example) - which started to become a significant share of the music market in the 80's, may explain some of the general trend".

The master thesis of Rudolf Ortner has a nice anecdote on that issue. Specifically, Ortner looks at "Beat it" by Michael Jackson, as of 1982, 1995, 2003 and 2009. As you can see, the volume has been pumped up markedly since 1982 (see figure below).

This is, however, just an anecdote.

(And this is also an anecdote on Michael Jackson and the "new sound", Michael Fremer as an expert witness involving Quincy Jones: https://www.analogplanet.com/conten...uincy-jones-vs-estate-michael-jackson-lawsuit ).

But there's still an answer to your question on loudness and genre. In a 2015 article, "THE MIR PERSPECTIVE ON THE EVOLUTION OF DYNAMICS IN MAINSTREAM MUSIC", Deruty and Pachet (http://ismir2015.uma.es/articles/136_Paper.pdf) argue that year is more important than genre:

"...we show that dynamics of mainstream music are more typical of a given year than they are of a given genre".

And further:

"It may seem counter-intuitive to conclude that dynamics are more dependent on trends than they are on genres. Indeed, genres such as Euro-Pop exhibit high micro- dynamics and low overall loudness, whereas other genres such as Trip-Hop are associated with low micro- dynamics and high overall loudness. However, the Euro- Pop genre is most represented in the 1970s and 1980s [34], at a period when music was produced to feature high microdynamics and low overall loudness [6]. Trip- Hop is mainly a mid-1990s trend [35], a moment when low microdynamics and high overall loudness were common in music production [6]. Conversely, all genres that span several decades follow the trend of the year of production".

Even more interestingly, Deruty and Pachet argue that the loudness war has made its peak, and is on retreat, possibly going back to the old values somewhere between 2017 and 2026. That's uplifting, isn't it?



-----------------

-----------------

 

[Floyd Toole]

By the way, I've heard that using several subwoofers was a good idea, although they all play the same mono signal...

Yes, multiple subs are a good idea. In simple rectangular rooms two or four identical subwoofers in the right locations, driven by the same signal, can attenuate some of the problematic room modes, leaving several locations for listeners to hear similar bass. Then EQ will satisfy multiple listeners. The next level of complication involves making measurements, running an optimization program and driving multiple loudspeakers with individually processed signals. Then the subs do not need to be identical, the locations are more flexible, and it works in non-rectangular rooms. Listeners in specified locations will then hear very similar bass. The room modes cease to be audible problems. It is all explained in Chapter 9.

 

[Guermantes]

It is my understanding 4 closely spaced cardioids at 90 degree angles are often used for the basic background ambient sound. With other sources mixed into that. The technique is common in live or sporting events done with surround sound. Typically referred to as the IRT cross. It is set up like double ORTF with wider spacing. Double MS is a coincident method which decodes to 4 super-cardioids. Either of these can be used for basic ambiance with two or three other mics up front for the focused sound up front. This can work nicely. Also allows you to alter the ratio of ambient hall sound and the musicians you wish to focus on.

IRT Cross:

Double M-S:

The latter looks very interesting -- I'm a fan of M-S stereo pairs.

 

[Blumlein 88]

IRT Cross:
View attachment 12787

Double M-S:
View attachment 12788

The latter looks very interesting -- I'm a fan of M-S stereo pairs.

I'm a fan of M-S too. I have a Lewitt TS640 which allows me to choose the pattern after the recording is done. Combined with a bi-directional microphone it will let me choose patterns, direction and number of channels after the fact. So I could have double MS and coincident 5 cardioids or others if I wish all from one three channel recording. Family illness has prevented me from using it much yet. Hopefully I'll soon be able to make some recordings this way.

 

[Theo]

But there's still an answer to your question on loudness and genre. In a 2015 article, "THE MIR PERSPECTIVE ON THE EVOLUTION OF DYNAMICS IN MAINSTREAM MUSIC", Deruty and Pachet (http://ismir2015.uma.es/articles/136_Paper.pdf) argue that year is more important than genre:

Thank you Svart (may I call you Svart?). Interesting paper, which covers most of the mainstream music genres. However, as detailed in Figure 2, most, if not all, samples are with amplified instruments and are "described as a contest between bands and record companies, in which music is engineered to be louder than the competition’s". What I would be interested in is the evolution of acoustic music recordings such as classical or live jazz which may not go through the loudness war as much as the 'Top 50' songs. Is the double bass section in classical music compressed and "enhanced" nowadays? Is the share of acoustic music in the statistics computed by Rudolf Ortner significant and how does it influence the result?

 

[Fitzcaraldo215]

I'm still not convinced that there isn't something special about stereo. My view of stereo is not that it is merely twice as good as mono, but more akin to the rendering of a hologram - in effect replacing a thousand 'channels' if you want to see it that way. Those 'channels' are all in front of the listener, however (not including BACCH and related ideas).

It would appear obvious that if stereo sounds so good in one 'quadrant' then with four speakers around the listener it should be possible to create an all-round 'hologram', but I presume it doesn't work that way! What works with two speakers ahead of the listener doesn't work front-to-back, and nor can the placement be unambiguous if a single impulse is rendered from more than two speakers.

I presume this is why multi-channel then goes for many channels and speakers around the listener - and then it looks to someone like me as though a science fiction 'sound field renderer' has been replaced by something that is much more of a gimmick. Maybe a very enjoyable gimmick, but nonetheless not as sophisticated or miraculous as stereo.

Mch is not a gimmick, I can assure you. I do not think Dr. Toole, among many others, would endorse and praise it if it were. And, that is true for music just as much as it is for videos.

I grant you that it has little benefit for many popular, multitrack music mixes made in dead studio acoustics other than to allow the panning of performers around you in "surround sound". Quite possibly, that is just a gimmick.

But, if you know and love the sound of live classical or acoustic music in a decent concert hall, Mch does that far better than stereo. Why? The mic array and Mch recording system capture more of the actual direct/reflected sound field heard naturally by listeners in the hall, including a much better approximation of 360 degree directional information. Stereo truncates that to just L and R, about 60 degrees. So, there is more capture and reproduction of much more information about the sound field we actually hear live in the hall at the performance.

On playback, all the wonderful holographic presentation of stereo that you love is still there up front in Mch. Why would that suddenly disappear? Only, it is better because Mch can also rely on the center channel and reduce its dependence on stereo's phantom imaging at the sensitive center of our directional hearing acuity, straight ahead. Dr. Toole also makes the point that this has significant positive effects on perceived frequency response in the center of the image. And, added to that improved frontal image is the natural enveloping ambiance of the hall reflections coming from the proper directions in 360 degree space.

I could go on and on, but I think you are looking at the issue from the wrong end of the telescope. It might help to get a deeper and more complete understanding before declaring it a gimmick. Hearing it properly set up with music might sometimes be difficult, but you might find that greatly improves your understanding.

Dr. Toole and others cited many of the old lame rationalizations, conspiracy theories, etc. offered by stubborn audiophiles about there being no need for stereo when mono was already so "good". Sound familiar?

 

[Cosmik]

Mch is not a gimmick, I can assure you. I do not think Dr. Toole, among many others, would endorse and praise it if it were. And, that is true for music just as much as it is for videos.

I grant you that it has little benefit for many popular, multitrack music mixes made in dead studio acoustics other than to allow the panning of performers around you in "surround sound". Quite possibly, that is just a gimmick.

But, if you know and love the sound of live classical or acoustic music in a decent concert hall, Mch does that far better than stereo. Why? The mic array and Mch recording system capture more of the actual direct/reflected sound field heard naturally by listeners in the hall, including a much better approximation of 360 degree directional information. Stereo truncates that to just L and R, about 60 degrees. So, there is more capture and reproduction of much more information about the sound field we actually hear live in the hall at the performance.

On playback, all the wonderful holographic presentation of stereo that you love is still there up front in Mch. Why would that suddenly disappear? Only, it is better because Mch can also rely on the center channel and reduce its dependence on stereo's phantom imaging at the sensitive center of our directional hearing acuity, straight ahead. Dr. Toole also makes the point that this has significant positive effects on perceived frequency response in the center of the image. And, added to that improved frontal image is the natural enveloping ambiance of the hall reflections coming from the proper directions in 360 degree space.

I could go on and on, but I think you are looking at the issue from the wrong end of the telescope. It might help to get a deeper and more complete understanding before declaring it a gimmick. Hearing it properly set up with music might sometimes be difficult, but you might find that greatly improves your understanding.

Dr. Toole and others cited many of the old lame rationalizations, conspiracy theories, etc. offered by stubborn audiophiles about there being no need for stereo when mono was already so "good". Sound familiar?

I'm still trying to find someone who has an opinion on the 'triangulation' aspect i.e. if you triangulate with two lines to find a location, the result is unambiguous: two lines cross at some position and that is that. If you introduce a third line, more information is provided but it is also more ambiguous with several possible 'solutions'.

Stereo (whatever the finer points of HRTF, etc.) provides a jewel-like, seemingly stable 'image' (if you don't move around too much). Is it so good because it comes from 'triangulating' only two channels?

As discussed above, there's no problem with piping some 'hall preset' to some surround speakers while keeping stereo at the front. It sounds great. But it's kind of obvious 'how it's done'. Stereo, to me, produces a greater result than the sum of the parts would suggest, and it just feels as though people are dismissing it out of hand because it only has two channels and more must be better. It is not obvious to me that multi-channel is necessarily a 'scientific' improvement on stereo for all situations - and might sometimes be worse. But I am open to persuasion.

 

[fredoamigo]

On playback, all the wonderful holographic presentation of stereo that you love is still there up front in Mch. Why would that suddenly disappear? Only, it is better because Mch can also rely on the center channel and reduce its dependence on stereo's phantom imaging at the sensitive center of our directional hearing acuity, straight ahead. Dr. Toole also makes the point that this has significant positive effects on perceived frequency response in the center of the image. And, added to that improved frontal image is the natural enveloping ambiance of the hall reflections coming from the proper directions in 360 degree space.[/QUOTE

then? why, for example, anthem doesn't use the central channel in music configuration?

AnthemLogic-Music (section 25)

https://www.anthemav.com/downloads/avm-mrx-english-manual.pdf

 

[Floyd Toole]

I'm still trying to find someone who has an opinion on the 'triangulation' aspect i.e. if you triangulate with two lines to find a location, the result is unambiguous: two lines cross at some position and that is that. If you introduce a third line, more information is provided but it is also more ambiguous with several possible 'solutions'.

Stereo (whatever the finer points of HRTF, etc.) provides a jewel-like, seemingly stable 'image' (if you don't move around too much). Is it so good because it comes from 'triangulating' only two channels?

As discussed above, there's no problem with piping some 'hall preset' to some surround speakers while keeping stereo at the front. It sounds great. But it's kind of obvious 'how it's done'. Stereo, to me, produces a greater result than the sum of the parts would suggest, and it just feels as though people are dismissing it out of hand because it only has two channels and more must be better. It is not obvious to me that multi-channel is necessarily a 'scientific' improvement on stereo for all situations - and might sometimes be worse. But I am open to persuasion.

I'm not going to engage in what seems to be a matter of personal preference - that always wins, for the best of reasons. However, there are no mysteries about how the binaural hearing system constructs a soundstage from two channels. It is not explained by geometric triangulation. Sound images perceived across the front soundstage are the result of mono left, mono right, and phantom images created by double-mono amplitude panned (by attenuators - a pan pot - or by directional coincident stereo mics). If the mics are spaced, there is an additional time difference factor. All phantom images perceived between the loudspeakers are the result of two sounds arriving at each ear, not one, and as a result there is an inherent comb-filtering issue, most obvious for a center image. The image shifting is illustrated in the first phase of the precedence effect shown in my book in Figure 7.23(a), and the comb filter is shown in Figure 7.2. Ambiance and "air" come from less correlated content in the two channels - either reflected sounds picked up by mics and mixed in or added by an electronic simulator. For stability of localization and for accurate timbre of the featured artist a center channel is advantageous, and this was known in the 1930s, but the limitations of the LP (only two modulated groove sides) prevented more than two channels from the outset, and the rest is history. It has absolutely nothing to do with the two-ears/two-channels relationship - except in binaural/headphone recordings.

The dominant missing ingredient in stereo is envelopment - the feeling of being in the venue with the performers. Delivering this does not require many extra channels. In fact, when examined in detail, it turns out that, for a single listener, adding only two more in the right locations can get one remarkably close to being surrounded by up to 24 channels/speakers. See Section 15.7.1. The present infatuation with "immersive" sound, employing many more channels is driven mainly by movies, and there the desired illusion is one of auditory objects that can be localized in many different well defined places. These are gee-whiz sound effects, not at all the uncorrelated spaciousness of envelopment, although that too is possible - with, as I said, many fewer channels. If one wishes to entertain more than one listener, more channels are advantageous, and this is the case in movies and in most home theaters.

The sounds that are delivered to the additional channels to create envelopment is the remaining choice, and one can either concoct a simulated sound field using a synthesizer, or extract the uncorrelated information from the recording itself and deliver it by an upmixer.

All that said, there is much pleasure to be gained from simple stereo - it is massively superior to mono, especially when it is well mixed. In my perusal of the recorded repertoire on TIDAL it is evident that stereo recordings are often amateurish, even for some well recognized performers.

 

[Floyd Toole]

Fredoamigo said: "then? why, for example, anthem doesn't use the central channel in music configuration?"

You'd have to ask them, but my guess is that it is because the recording was not made using a center channel and there is no center channel signal feed. Upmixers can "derive" a center channel, and that will help with stabilizing localization, but because the mix was done listening to a phantom center, the timbre chosen by the recording engineer may not be accurately reproduced. If you want the stability of a derived center channel switch to Anthem Cinema, or any of the Dolby and DTS options which add more surround information as well - they are different because the intent is to satisfy multiple listeners.

 

[Fitzcaraldo215]

I'm still trying to find someone who has an opinion on the 'triangulation' aspect i.e. if you triangulate with two lines to find a location, the result is unambiguous: two lines cross at some position and that is that. If you introduce a third line, more information is provided but it is also more ambiguous with several possible 'solutions'.

Stereo (whatever the finer points of HRTF, etc.) provides a jewel-like, seemingly stable 'image' (if you don't move around too much). Is it so good because it comes from 'triangulating' only two channels?

As discussed above, there's no problem with piping some 'hall preset' to some surround speakers while keeping stereo at the front. It sounds great. But it's kind of obvious 'how it's done'. Stereo, to me, produces a greater result than the sum of the parts would suggest, and it just feels as though people are dismissing it out of hand because it only has two channels and more must be better. It is not obvious to me that multi-channel is necessarily a 'scientific' improvement on stereo for all situations - and might sometimes be worse. But I am open to persuasion.

I don't hear this triangulation issue at all in comparing stereo with 5.0/.1 Mch, and I don't hear any advantage in the frontal soundstage offered by stereo. In fact, I think Mch delivers a somewhat - noticeably but not hugely - better frontal soundstage aided by the center channel. This is mostly in perceived image dimensions of depth and width, but also in apparent clarity and resolution of images near the center of the soundstage. Stereo is entirely a phantom image in the center soundstage area, and phantom images are known to be less accurate and precise.

Try it yourself and see rather than speculate. Toole, I believe, explains much of why this is so In his book.

There are some remastered old Mercurys and RCAs originally recorded in 3-channel. People I know who have heard them prefer the 3-channel playback over 2-channel from the same SACD. These discs avoid possible confusion by surround immersive effects and focus our attention mainly just on the frontal soundstage.

 

[Cosmik]

I'm not going to engage in what seems to be a matter of personal preference - that always wins, for the best of reasons. However, there are no mysteries about how the binaural hearing system constructs a soundstage from two channels. It is not explained by geometric triangulation. Sound images perceived across the front soundstage are the result of mono left, mono right, and phantom images created by double-mono amplitude panned (by attenuators - a pan pot - or by directional coincident stereo mics). All phantom images perceived between the loudspeakers are the result of two sounds arriving at each ear, not one, and as a result there is an inherent comb-filtering issue, most obvious for a center image. The image shifting is illustrated in the first phase of the precedence effect shown in my book in Figure 7.23(a), and the comb filter is shown in Figure 7.2. Ambiance and "air" come from less correlated content in the two channels - either reflected sounds picked up by mics and mixed in or added by an electronic simulator. For stability of localization and for accurate timbre of the featured artist a center channel is advantageous, and this was known in the 1930s, but the limitations of the LP (only two modulated groove sides) prevented more than two channels from the outset, and the rest is history. It has absolutely nothing to do with the two-ears/two-channels relationship - except in binaural/headphone recordings.

The dominant missing ingredient in stereo is envelopment - the feeling of being in the venue with the performers. Delivering this does not require many extra channels. In fact, when examined in detail, it turns out that, for a single listener, adding only two more in the right locations can get one remarkably close to being surrounded by up to 24 channels/speakers. See Section 15.7.1. The present infatuation with "immersive" sound, employing many more channels is driven mainly by movies, and there the desired illusion is one of auditory objects that can be localized in many different well defined places. These are gee-whiz sound effects, not at all the uncorrelated spaciousness of envelopment, although that too is possible - with, as I said, many fewer channels. If one wishes to entertain more than one listener, more channels are advantageous, and this is the case in movies and in most home theaters.

The sounds that are delivered to the additional channels to create envelopment is the remaining choice, and one can either concoct a simulated sound field using a synthesizer, or extract the uncorrelated information from the recording itself and deliver it by an upmixer.

All that said, there is much pleasure to be gained from simple stereo - it is massively superior to mono, especially when it is well mixed. In my perusal of the recorded repertoire on TIDAL it is evident that stereo recordings are often amateurish, even for some well recognized performers.

I'm obviously not explaining myself very well re. 'triangulation' - maybe that's the wrong word. I'm referring to identifying the location of an emitter of an impulse based on the relative timing of the impulse's arrival at two spaced detectors.

If there is an 'impulse' played in stereo, the imaging seems to work remarkably well. It's as though the listener is hearing a simulation of a source from a position in space that remains approximately constant with minor changes of head position. Perhaps it shouldn't work as well as it does! . It's limited to being in front of you (no doubt an artefact of the leakage), but it is nevertheless a beautifully stable, 'refined' sound with jewel-like imaging, and the room does provide some all-around 'ambience' that cements the illusion.

But what is the effect of introducing a third source derived from that same 'impulse'? To me, that seems like a fundamental difference that could introduce ambiguity in the locating of the source. With stereo, the positions of the speakers and the listener are not critical: you will get a single 'solution' for the apparent position of the emitter (notwithstanding the cross-channel leakage that doesn't seem nearly as detrimental as you would imagine). Introduce the third speaker and suddenly there is no solution to the location unless you happen to have super-critical positioning of listener and speakers.

Is this a real concern?

 

[Floyd Toole]

Introduce the third speaker and suddenly there is no solution to the location unless you happen to have super-critical positioning of listener and speakers.

Is this a real concern?

I think not, because of the linear superposition of the sound fields: (1) the original stereo and (2) a center channel. Only if the exact same sound is emitted from all three would there be a possible problem, and this is highly unlikely, and indeed undesirable. Two ears and a brain would have no problem sorting it out. Think of it as the superposition of the sound of a choir (where "phantom" images abound contributing to the "chorus" effect), and the sound of a soloist who is separately localized.

That said, it occasionally happens in multichannel video concert recordings, where what should be a center image is sent to all three front speakers.