Can Loudspeakers Accurately Reproduce The Sound Of Real Instruments...and Do You Care?

[Cosmik]

Symphonic stuff doesn't fare well imho, even if you can reproduce the SPL and have very good imaging. Dr Bose wasn't totally wrong when he decided to focus on reflections in the 901s Two front facing sources won't reproduce the concert hall imo.

I have attended many wonderful virtual symphonic concerts thanks to those two front facing sources - and I do mean that the experience has been on a par with the real thing. Big speakers have no trouble with the required SPL and richness (since when has a seat in the audience of a symphonic concert been at hearing damage levels, anyway?). And thanks to the wonder of stereo, the two speakers render a complete, stable, detailed panorama that subtends a similar angle to the real thing.

I think audiophiles have lost the plot a bit on this. Maybe their boutique-ish systems are simply not capable of creating that seamless panorama, or they are listening *to* their systems rather than through them, or they are incapable of suspending their disbelief at what is possible with two speakers. For example, someone earlier mentioned the acoustic crosstalk from speakers to the 'wrong ear' as a problem, when in fact it is the crosstalk that creates the image. If you approach it with that 'problem' in mind, maybe you can't let yourself believe what you are hearing. Audiophiles are burdened with knowing (some of) how the music is being produced. They need to let go a bit .

 

[Shadrach]

I have attended many wonderful virtual symphonic concerts thanks to those two front facing sources - and I do mean that the experience has been on a par with the real thing. Big speakers have no trouble with the required SPL and richness (since when has a seat in the audience of a symphonic concert been at hearing damage levels, anyway?). And thanks to the wonder of stereo, the two speakers render a complete, stable, detailed panorama that subtends a similar angle to the real thing.

I think audiophiles have lost the plot a bit on this. Maybe their boutique-ish systems are simply not capable of creating that seamless panorama, or they are listening *to* their systems rather than through them, or they are incapable of suspending their disbelief at what is possible with two speakers. For example, someone earlier mentioned the acoustic crosstalk from speakers to the 'wrong ear' as a problem, when in fact it is the crosstalk that creates the image. If you approach it with that 'problem' in mind, maybe you can't let yourself believe what you are hearing. Audiophiles are burdened with knowing (some of) how the music is being produced. They need to let go a bit .

I'll be round with my tenor sax. Make a dental appointment.

 

[Xulonn]

How could double bass or cello music not sound fantastic and realistic on these loudspeakers? /snark

 

[svart-hvitt]

This is a hopeless experiment that simply can not work, and this is why:

- The recording does not capture the sound emitted from the instrument, because the mic receives sound from one direction only, while the radiation pattern of the instrument will be highly frequency dependent and closer to omni in parts of the frequency band.
- The speaker does not have the same radiation pattern as the instrument, which will induce faults in emitted energy for different frequency bands and different frequency response from reflections.

To make this work, you need to measure the instruments from ALL angles, and the reproduce using a speaker with radiation pattern similar to that of the instrument. Which means one speaker for each particular instrument.

But this has no relevance to what we are trying to do at home - reproduce the recording of an event, a recording which includes the reflections form the concert hall. More or less successful, depending on the recording/production and the reproduction system - speakers/room.

@Kvalsvoll, Pätynen defende his PhD, a work during which professor Tapio Lokki was his instructor. Lokki followed up the PhD with an ISRA article a couple of years later: http://www.caa-aca.ca/conferences/isra2013/proceedings/Papers/P040.pdf

It would be suprising if you from armchair’s distance were in a position to write off both the whole PhD idea and later work based on that idea. @Floyd Toole commented previously on Lokki and did not at all write him off as a “hopeless” experimenter. When you use strong words, I expect strong arguments.

With regard to your mentioning of “omni”, this word (omni) is mentioned 37 times in the PhD thesis. So this is not a point lost on the author. In Lokki (2013), the author writes:

“2 OBJECTIVE MEASUREMENTS OF CONCERT HALLS
As the subjective comparison of concert halls is difficult and often biased by matters of taste, re- searchers have tried to invent objective ways to measure some features of the acoustics. This work has lead to the definition of acoustic parameters described in the international standard ISO3382- 1:200927, illustrated in Figure 1. The standard requires that room impulse responses should be measured with an omnidirectional loudspeaker from a few source positions on the stage to 6-10 receiver positions in the audience area. The capturing microphone has omni or figure-of-eight directivity. The parameters are computed from the sound energy decays at different frequency bands.
The ISO3382-1:2009 standard has been criticized from many angles28,29. For example, the algo- rithms to compute the parameters are imprecise, the applied frequency range is inadequate, and a single omnidirectional source does not correspond to real orchestra, which, in reality has dozens of sources with varying directivity characteristics. Moreover, averaging results over several receiver positions hides information as the parameters vary at different distances quite a lot. However, it is generally accepted that some of the standard parameters correlate quite well with subjective perceptions, e.g. Strength (G) with loudness and early decay time (EDT) with reverberance. In contrast, other perceptually relevant factors, e.g., intimacy, have no corresponding objective pa- rameter and values for parameters that correlate with subjective preference judgements are very cumbersome to define.
In order to overcome the inherent simplifications in standard ISO parameters, we have recently taken a novel approach to measure concert hall acoustics, illustrated in Figure 2. The sound source is the same loudspeaker orchestra used to simulate symphony orchestra for subjective studies. The directivities of the loudspeakers differ from the directivities of real instruments, but we have tried to minimize the possible errors by choosing appropriate loudspeakers18. The spatial impulse responses for objective analysis from all source positions are captured with a microphone array, currently with six omnidirectional microphones. We have developed a spatial decompo- sition technique (SDM)24 to analyze the spatial distribution of sound energy from every source loudspeaker. Moreover, new time-frequency and spatiotemporal visualization techniques allow examination of sound energy levels in many dimensions: time, frequency, and space (azimuth and elevation). Our recent article30 proposes initial ideas to link the development of the spatial sound field over time to the plan and section drawings of the measured concert halls (Figure 2)”.

So your remark on a “hopeless experiment” seems rushed considering the fact that the researchers have thought both long and well about radiation.

Having said that, my main point was that figures 4.2 to 4.5 (see my post #35: https://www.audiosciencereview.com/...ments-and-do-you-care.7422/page-2#post-171823) from Pätynen’s PhD thesis document a gap between an instrument’s power response and the recorded power response through a speaker. The systematic bias in the speaker response (too little high frequency power?) is surprising, at least to me, and nobody has come up with an explanation for the gap.

 

[MattHooper]

I think audiophiles have lost the plot a bit on this. Maybe their boutique-ish systems are simply not capable of creating that seamless panorama, or they are listening *to* their systems rather than through them, or they are incapable of suspending their disbelief at what is possible with two speakers. For example, someone earlier mentioned the acoustic crosstalk from speakers to the 'wrong ear' as a problem, when in fact it is the crosstalk that creates the image. If you approach it with that 'problem' in mind, maybe you can't let yourself believe what you are hearing. Audiophiles are burdened with knowing (some of) how the music is being produced. They need to let go a bit .

I dunno. I've seen plenty of audiophiles who are quite pleased with how their systems present symphonic music.

Personally, I've had a few speakers that left me happily impressed: with eyes closed, the spatial sensation of being at the symphony has been quite impressive, even when it has meant the sensation of being seated further back from the orchestra.

I agree that there is an interaction between the mental attitude of the listener and the speaker system. The more one "let's go" the easier it is to accept an illusion. And that brings up something that's always been something of a conundrum about audiophile listening. Many audiophiles say they ultimately seek to "lose themselves" in the music. Yet the more one loses themselves in the music, the less attention one is actually paying to the sound of the system. If you aren't going to take some note of the *sound* of your system, why pay all that money for it?

I can lose myself in music easily from any number of crappy audio sources. With my high end system not only do I enjoy the music itself, but the sonic quality itself rewards attention in a way it doesn't with a crappy audio system. When I have that balance right, the entire experience of listening to music on my high end system is richer and more compelling.

 

[Kvalsvoll]

@Kvalsvoll, Pätynen defende his PhD, a work during which professor Tapio Lokki was his instructor. Lokki followed up the PhD with an ISRA article a couple of years later: http://www.caa-aca.ca/conferences/isra2013/proceedings/Papers/P040.pdf

It would be suprising if you from armchair’s distance were in a position to write off both the whole PhD idea and later work based on that idea. @Floyd Toole commented previously on Lokki and did not at all write him off as a “hopeless” experimenter. When you use strong words, I expect strong arguments.

With regard to your mentioning of “omni”, this word (omni) is mentioned 37 times in the PhD thesis. So this is not a point lost on the author. In Lokki (2013), the author writes:

“2 OBJECTIVE MEASUREMENTS OF CONCERT HALLS
As the subjective comparison of concert halls is difficult and often biased by matters of taste, re- searchers have tried to invent objective ways to measure some features of the acoustics. This work has lead to the definition of acoustic parameters described in the international standard ISO3382- 1:200927, illustrated in Figure 1. The standard requires that room impulse responses should be measured with an omnidirectional loudspeaker from a few source positions on the stage to 6-10 receiver positions in the audience area. The capturing microphone has omni or figure-of-eight directivity. The parameters are computed from the sound energy decays at different frequency bands.
The ISO3382-1:2009 standard has been criticized from many angles28,29. For example, the algo- rithms to compute the parameters are imprecise, the applied frequency range is inadequate, and a single omnidirectional source does not correspond to real orchestra, which, in reality has dozens of sources with varying directivity characteristics. Moreover, averaging results over several receiver positions hides information as the parameters vary at different distances quite a lot. However, it is generally accepted that some of the standard parameters correlate quite well with subjective perceptions, e.g. Strength (G) with loudness and early decay time (EDT) with reverberance. In contrast, other perceptually relevant factors, e.g., intimacy, have no corresponding objective pa- rameter and values for parameters that correlate with subjective preference judgements are very cumbersome to define.
In order to overcome the inherent simplifications in standard ISO parameters, we have recently taken a novel approach to measure concert hall acoustics, illustrated in Figure 2. The sound source is the same loudspeaker orchestra used to simulate symphony orchestra for subjective studies. The directivities of the loudspeakers differ from the directivities of real instruments, but we have tried to minimize the possible errors by choosing appropriate loudspeakers18. The spatial impulse responses for objective analysis from all source positions are captured with a microphone array, currently with six omnidirectional microphones. We have developed a spatial decompo- sition technique (SDM)24 to analyze the spatial distribution of sound energy from every source loudspeaker. Moreover, new time-frequency and spatiotemporal visualization techniques allow examination of sound energy levels in many dimensions: time, frequency, and space (azimuth and elevation). Our recent article30 proposes initial ideas to link the development of the spatial sound field over time to the plan and section drawings of the measured concert halls (Figure 2)”.

So your remark on a “hopeless experiment” seems rushed considering the fact that the researchers have thought both long and well about radiation.

Having said that, my main point was that figures 4.2 to 4.5 (see my post #35: https://www.audiosciencereview.com/...ments-and-do-you-care.7422/page-2#post-171823) from Pätynen’s PhD thesis document a gap between an instrument’s power response and the recorded power response through a speaker. The systematic bias in the speaker response (too little high frequency power?) is surprising, at least to me, and nobody has come up with an explanation for the gap.

The quote above about measurements is related to how acoustic properties of a concert hall should be measured, while trying to use speakers to replicate instruments in the same hall is not the same.

The experiment must also be seen in context. For its intended purpose, the chosen solution may have been sufficient, it is not necessarily so that the intended audience would require instruments to sound exactly similar to the real ones, to be able to enjoy a performance.

The measured differences in power spectrum is caused by both the recording method, and by the reproduction using speakers with a different radiation pattern than the real instruments. This causes too little energy at higher frequencies, for most of the instruments. This could be predicted, from understanding what happens in the recording, and from the radiation pattern of the speakers. The measurements only confirm these facts.

 

[Bjorn]

Obviously there's the challenge with the recording and optimal recording techniques and the speakers need to be superb, but oivavoi touched on something that doesn't seem to be very popular to debate here and that is the acoustics. You cannot take this part out as it has a huge effect one areas lik the timbre, localization, resolution, spaciousness, etc.. You have to emulate the concert hall acoustics to some degree and that's actually possible with a dedicated room of decent size. It was studied over decades.

 

[Bjorn]

Actually, most instruments are quite directional at higher frequencies.

 

[svart-hvitt]

The quote above about measurements is related to how acoustic properties of a concert hall should be measured, while trying to use speakers to replicate instruments in the same hall is not the same.

The experiment must also be seen in context. For its intended purpose, the chosen solution may have been sufficient, it is not necessarily so that the intended audience would require instruments to sound exactly similar to the real ones, to be able to enjoy a performance.

The measured differences in power spectrum is caused by both the recording method, and by the reproduction using speakers with a different radiation pattern than the real instruments. This causes too little energy at higher frequencies, for most of the instruments. This could be predicted, from understanding what happens in the recording, and from the radiation pattern of the speakers. The measurements only confirm these facts.

I don’t see why the initial original recording method plays a role here. The instrument’s sound is captured by the microphone and recorded to file. Fidelity between microphone and file is guaranteed.

Directivity of speaker in reproduction plays a role, however. Pätynen wrote:

«The reproduced trumpet and trombone power re- sponses are particularly close to the real instrument power responses (see Fig. 4.3). Otherwise, the power response of the instruments is character- ized reasonably well by a frequency response in a single direction. The slight low-pass behavior observed e.g. with the woodwind instruments and the cello suggests that the applied loudspeaker is too directional».

 

[Blumlein 88]

I've found doing close miking either in a dead room or a large space, playing back that one microphone feed over one speaker makes for a big jump in realism. I've done simple experiments where I had 3 to 5 musicians with each having their own mike. Then playback over 3 to 5 speakers so each musician gets a speaker.

For dead room recordings there is a strong sense of the musician being right here where you are. If the recording is in a large space, that does seem to add some of the sound of the longer delayed reflections and it is more of a you are there effect. Both versions are more 'live' than when you do virtual images spread out to more than one speaker.

This is so despite all the issues like the speaker directivity not being what the instrument/musician would have been. Like the microphone only picking up sound in one spot in space though close to the sound source. etc etc. Reasons I would expect this wouldn't work as well as it does. Or another one which is you can use two microphones close miking in two different spots. They sound somewhat different, but either applied to a single speaker has a good realism quotient. Can they both be right? Does one have to be wrong? I would venture they are likely both wrong. But they sound pretty right. Then again, when listening live to say a guitar, you hear the first arrival sound from any different angle and it always seems like a guitar.

 

[MattHooper]

Great input Blumlein 88, thanks!

Though I have been in a band or two and done some studio recording, the vast majority of my recording has been for my job as a sound effects designer/editor. Sometimes I'm going for a natural recording, other times it's intended to be manipulated. But though I spend a lot of time very precisely manipulating sounds to match (be it a baby's voice or even the "air" of room), I don't have the specific goal of the type of timbrel exactness I'm talking about - the selection of the sound system by the user sort of puts that out of my hands.

svart-hvitt,

Thanks very much for those very interesting graphs!!!

 

[Kvalsvoll]

I don’t see why the initial original recording method plays a role here. The instrument’s sound is captured by the microphone and recorded to file. Fidelity between microphone and file is guaranteed.

Directivity of speaker in reproduction plays a role, however. Pätynen wrote:

«The reproduced trumpet and trombone power re- sponses are particularly close to the real instrument power responses (see Fig. 4.3). Otherwise, the power response of the instruments is character- ized reasonably well by a frequency response in a single direction. The slight low-pass behavior observed e.g. with the woodwind instruments and the cello suggests that the applied loudspeaker is too directional».

After reading parts of this thesis (A virtual symphony orchestra for studies on concert hall acoustics Jukka Pätynen ), which is actually quite interesting, my previous comments should be disregarded, as they were based on assumptions I did, which were wrong, about how this experiment was done.

The recordings were made using 22 microphones, not 1. Instrument directivity, method of recording and post-processing of the recorded signals are certainly considered and taken into account for. The recordings contain all sound from the instruments.

The loudspeaker is not one or two speakers, it is 24 speakers.

The deviations in power response is also - at least partially - explained. It is mentioned that loudspeaker directivity is the cause (Your (svart-hvitt) quote above). For the string sections, they applied additional speakers to give a wider radiation pattern for a better match to the real instrument.

The purpose of all this is not to fire all the musicians, it is to find better ways to evaluate acoustic properties of concert halls.

The thesis also investigates directivity of instruments, and has lots of information on that.

Fortunately, it does not require 24 speakers in our room to reproduce a recording. Playing back a recording is not the same as trying to reproduce an instrument placed in the room.

 

[Juhazi]

This discussion makes me think, why most pro musicians are not into hifi. They hear natural sound daily and know how that sound varies in different rooms and halls. There is no specific sound to an instrument or voice, it always varies! This is different from choosing your personal instrument by the sound and playability of it.

So, musicians and 98% of all people listen to what is played and how it's played, in terms of arrangement, touch, timing, rhythm, intonation,etc. The rest 2% are people interested in hifi reproduction systems...

 

[svart-hvitt]

This discussion makes me think, why most pro musicians are not into hifi. They hear natural sound daily and know how that sound varies in different rooms and halls. There is no specific sound to an instrument or voice, it always varies! This is different from choosing your personal instrument by the sound and playability of it.

So, musicians and 98% of all people listen to what is played and how it's played, in terms of arrangement, touch, timing, rhythm, intonation,etc. The rest 2% are people interested in hifi reproduction systems...

My avatar:

The famous pipe. How people reproached me for it! And yet, could you stuff my pipe? No, it's just a representation, is it not? So if I had written on my picture 'This is a pipe', I'd have been lying!

— René Magritte

Background: https://en.m.wikipedia.org/wiki/The_Treachery_of_Images

 

[BillG]

This discussion makes me think, why most pro musicians are not into hifi. They hear natural sound daily and know how that sound varies in different rooms and halls. There is no specific sound to an instrument or voice, it always varies! This is different from choosing your personal instrument by the sound and playability of it.

So, musicians and 98% of all people listen to what is played and how it's played, in terms of arrangement, touch, timing, rhythm, intonation,etc. The rest 2% are people interested in hifi reproduction systems...

As a melophile, I go for a home system that can sound as close in clarity to the best nightclub systems I've heard - we've great nightlife here in Auckland, New Zealand, with a number of clubs with high caliber sound systems. Granted I don't need all the SPL that they can pump out, but if mine is as clear and distortion free as those setups, I'm happy...

 

[Shadrach]

I used to use this model to explain to Electrical Engineers and Recording Engineers some of the problems concerning acoustics. The model has its limitations, the ability of ‘sound’ to travel through materials that light wont, decay times, fundamentals and harmonics translating to frequencies of light, for examples.
The first thing is to change the vocabulary used; this is a similar problem when using the term ‘music’ when in fact we are talking about signal processing.
So, instead of sound, lets use the more accurate term, pressure wave.
Imagine an orchestra set up in a concert hall. The musicians are not usually placed in a line and not all the musicians face the same direction. In effect you have a three dimensional scattering of pressure wave sources.
Now imagine you replace the instruments with torches. These torches have the ability to focus light with the aid of a lens from a beam to a 360 degree flood
Lets ascribe frequency to this ability.
Now modify these torches so that they are adjustable with regard to light intensity throughout their focal range.
Ascribe volume to this ability.
Each note on a musical score is then translated to a fixed frequency and intensity (frequency and volume).
The orchestra plays and you have a light show, which with a little imagination can be viewed as the interactions of all the pressure waves should the orchestra be playing instruments.
Cut a plane on any axis and you have a two dimensional representation of all the interactions of the light waves at their various frequencies and volumes.
Now carry out the same process for a stereo system playing the same part of the score used above and a plane cut at the same event point.
Make this ‘cut’ transparent and overlay it on the plane cut from the orchestra.
For an accurate representation of the orchestra light show the stereo light show overlay should be identical.
Place yourself as the observer in any position on the plane.
While the model is far from perfect wave properties have enough similarities to make the model useful.

 

[svart-hvitt]

I used to use this model to explain to Electrical Engineers and Recording Engineers some of the problems concerning acoustics. The model has its limitations, the ability of ‘sound’ to travel through materials that light wont, decay times, fundamentals and harmonics translating to frequencies of light, for examples.
The first thing is to change the vocabulary used; this is a similar problem when using the term ‘music’ when in fact we are talking about signal processing.
So, instead of sound, lets use the more accurate term, pressure wave.
Imagine an orchestra set up in a concert hall. The musicians are not usually placed in a line and not all the musicians face the same direction. In effect you have a three dimensional scattering of pressure wave sources.
Now imagine you replace the instruments with torches. These torches have the ability to focus light with the aid of a lens from a beam to a 360 degree flood
Lets ascribe frequency to this ability.
Now modify these torches so that they are adjustable with regard to light intensity throughout their focal range.
Ascribe volume to this ability.
Each note on a musical score is then translated to a fixed frequency and intensity (frequency and volume).
The orchestra plays and you have a light show, which with a little imagination can be viewed as the interactions of all the pressure waves should the orchestra be playing instruments.
Cut a plane on any axis and you have a two dimensional representation of all the interactions of the light waves at their various frequencies and volumes.
Now carry out the same process for a stereo system playing the same part of the score used above and a plane cut at the same event point.
Make this ‘cut’ transparent and overlay it on the plane cut from the orchestra.
For an accurate representation of the orchestra light show the stereo light show overlay should be identical.
Place yourself as the observer in any position on the plane.
While the model is far from perfect wave properties have enough similarities to make the model useful.

When discussing audio on a somewhat abstract level, talking about music can do more harm to the discussion than good. Many want to drag the discussion in the direction where emotional response is all that matters, while audio may have many other areas of interest than emotional arousal.

Speakers are sound machines that may or may not be used to reproduce the sound of music. Speakers are tools, and tools have different applications depending on use. Dental tools differ from coal mining tools; the one is not better than the other without defining one’s needs beforehand.

 

[Cosmik]

I used to use this model to explain to Electrical Engineers and Recording Engineers some of the problems concerning acoustics. The model has its limitations, the ability of ‘sound’ to travel through materials that light wont, decay times, fundamentals and harmonics translating to frequencies of light, for examples.
The first thing is to change the vocabulary used; this is a similar problem when using the term ‘music’ when in fact we are talking about signal processing.
So, instead of sound, lets use the more accurate term, pressure wave.
Imagine an orchestra set up in a concert hall. The musicians are not usually placed in a line and not all the musicians face the same direction. In effect you have a three dimensional scattering of pressure wave sources.
Now imagine you replace the instruments with torches. These torches have the ability to focus light with the aid of a lens from a beam to a 360 degree flood
Lets ascribe frequency to this ability.
Now modify these torches so that they are adjustable with regard to light intensity throughout their focal range.
Ascribe volume to this ability.
Each note on a musical score is then translated to a fixed frequency and intensity (frequency and volume).
The orchestra plays and you have a light show, which with a little imagination can be viewed as the interactions of all the pressure waves should the orchestra be playing instruments.
Cut a plane on any axis and you have a two dimensional representation of all the interactions of the light waves at their various frequencies and volumes.
Now carry out the same process for a stereo system playing the same part of the score used above and a plane cut at the same event point.
Make this ‘cut’ transparent and overlay it on the plane cut from the orchestra.
For an accurate representation of the orchestra light show the stereo light show overlay should be identical.
Place yourself as the observer in any position on the plane.
While the model is far from perfect wave properties have enough similarities to make the model useful.

I don't see the above as 100% relevant audio. The stereo recording is not attempting to capture the orchestra and concert hall (nice though that might be); it is attempting to capture/re-create the sound as it appears at two points and then render it in such a way that the listener experiences something similar to the person in the best seat in the house.

We are the beneficiaries of several serendipitous factors:

1. The technology to reproduce audio is compatible with the lifestyles and incomes of non-billionaires
2. Intentionally or otherwise, Mr. Blumlein stumbled across an idea whose effectiveness (and, I suspect, perfection) still hasn't been experienced by 99% of people. Not only does it work for someone sitting perfectly still, but it 'hangs together' with head turning or even walking about. Combined with the listening room acoustics this converts a static image into one that has a dynamic acoustic element. Subconsciously, we register this and it makes the experience even richer.
3. Direct radiating speakers' wide, benign dispersion characteristics and lack of coloration are compatible with audio reproduction in a normal room. Not only does this mean that they are easy to make and neutral in sound, but that they engage with the room in a similar way to the listeners' own voices, creating a two-way blend between the recording and the listening venues. There is no abrupt difference between the way the room responds dynamically to the recording and the way it responds to the listeners' voices (as there would be if the speakers were super-directional for example).
4. Stereo fools our hearing perfectly without any unpleasant side effects
5. We 'hear through the room'; registering both it and the source but not confusing the two - we *never* think the room is changing size, shape and materials if we turn 'room correction' on and off!

We get a plausible version of the event. People who are obsessed with accuracy (to criteria that someone has spun out of thin air) are on a hiding to nothing. Simple neutrality at the speakers will give you the plausible experience due to those fortunate factors above. Pursuing accuracy towards made-up criteria (e.g. the mic & FFT is a model of human hearing and the results must be altered to some made-up target curve by all means possible) is the way to destroy the plausibility.

 

[Shadrach]

I don't see the above as 100% relevant audio. The stereo recording is not attempting to capture the orchestra and concert hall (nice though that might be); it is attempting to capture/re-create the sound as it appears at two points and then render it in such a way that the listener experiences something similar to the person in the best seat in the house.

We are the beneficiaries of several serendipitous factors:

1. The technology to reproduce audio is compatible with the lifestyles and incomes of non-billionaires
2. Intentionally or otherwise, Mr. Blumlein stumbled across an idea whose effectiveness (and, I suspect, perfection) still hasn't been experienced by 99% of people. Not only does it work for someone sitting perfectly still, but it 'hangs together' with head turning or even walking about. Combined with the listening room acoustics this converts a static image into one that has a dynamic acoustic element. Subconsciously, we register this and it makes the experience even richer.
3. Direct radiating speakers' wide, benign dispersion characteristics and lack of coloration are compatible with audio reproduction in a normal room. Not only does this mean that they are easy to make and neutral in sound, but that they engage with the room in a similar way to the listeners' own voices, creating a two-way blend between the recording and the listening venues. There is no abrupt difference between the way the room responds dynamically to the recording and the way it responds to the listeners' voices (as there would be if the speakers were super-directional for example).
4. Stereo fools our hearing perfectly without any unpleasant side effects
5. We 'hear through the room'; registering both it and the source but not confusing the two - we *never* think the room is changing size, shape and materials if we turn 'room correction' on and off!

We get a plausible version of the event. People who are obsessed with accuracy (to criteria that someone has spun out of thin air) are on a hiding to nothing. Simple neutrality at the speakers will give you the plausible experience due to those fortunate factors above. Pursuing accuracy towards made-up criteria (e.g. the mic & FFT is a model of human hearing and the results must be altered to some made-up target curve by all means possible) is the way to destroy the plausibility.

I can't find a single point in the above that I agree with.
I was attempting to answer, or at least demonstrate, some of the problems involved in the first part of the question which is;
Can Loudspeakers Accurately Reproduce The Sound Of Real Instruments...and Do You Care?
I'm not sure which part of the question you are responding to; maybe the last bit.
I'm just going to take one quote from your post which seems to be a summary of your points.
"We get a plausible version of the event."
You may, I don't. I get entertained.

 

[Cosmik]

I can't find a single point in the above that I agree with.
I was attempting to answer, or at least demonstrate, some of the problems involved in the first part of the question which is;
Can Loudspeakers Accurately Reproduce The Sound Of Real Instruments...and Do You Care?
I'm not sure which part of the question you are responding to; maybe the last bit.
I'm just going to take one quote from your post which seems to be a summary of your points.
"We get a plausible version of the event."
You may, I don't. I get entertained.

Well the meaning of the question is open to interpretation. My interpretation is that it refers to loudspeakers in the context of how they are actually used, not some other use for them such as replacing the individual pipes and drums of a fairground organ - entertaining though such a contraption might be.

I am sorry that you can't suspend your disbelief when listening to recordings of music.