A Broad Discussion of Speakers with Major Audio Luminaries

[Penelinfi]

A small transducer is able to reproduce higher frequencies and these are "faster" than lower frequencies - that is all there is to it. Within a given bandwidth, all loudspeakers are equally "fast". There are no "fast" woofers because the "speed" is determined by the highest frequency it reproduces. If it is bass managed, that is typically 80 Hz. The room resonances are the "slow" part, and unless these are controlled, bass will be "slow", as MaxwellsEq said. Science is different from subjective reviewing - facts are involved.

Scientific facts also include moving a further distance in the same amount of time needs higher speed, like small woofer vs big woofer for the same input or output and their various efficiencies. I'd hope people on this forum would see that instead of blindly upvoting without thinking; regurgitation. In fact more likely to thumbs up only due to the fact that their idol figure wrote something.

Many smaller woofers have softer suspension , lower efficiency, more compensation to reproduce bass frequencies, leading to more excursion. More distance travelled, same time, same output volume.

Whether you can hear this still remains to be seen. Maybe someone can investigate how air responds with large surface area less movement vs small surface area more movement.

 

[Mart68]

Doubling down against the odds. Good luck!

 

[amirm]

Scientific facts also include moving a further distance in the same amount of time needs higher speed, like small woofer vs big woofer for the same input or output and their various efficiencies. I'd hope people on this forum would see that instead of blindly upvoting without thinking; regurgitation. In fact more likely to thumbs up only due to the fact that their idol figure wrote something.

You don't hear that "speed." You hear pressure waves. And those waves must follow simply linear relationship between frequency, wavelength and hence, velocity (which is a simple function of the former two).

Think of it this way. Have a driver playing at very faint volume where you can't even see the cone moving. Then crank it up where it is playing at is maximum excursion. You think the sound changes because the cone has farther to travel in the second case?

 

[MaxwellsEq]

More distance travelled, same time, same output volume.

Ah, I see. You are describing the maximum velocity of the coil past the magnetic poles (in a conventional driver) which is at the zero crossing point for symmetric waves. This won't have an impact on sound assuming the driver is operating linearly. The person I was responding to was talking about "fast loudspeakers" as perceived in a room some distance from the driver. This is a different matter.

 

[thewas]

Well designed larger drivers in their well chosen passband reproduce the same input signals exactly the same well as designed smaller ones, otherwise this would be measured as higher distortion (linear and non-linear) which is the deviation to the input signal. This shows that "fast loudspeakers" is one of the unfortunately widely spread typical audiophile urban myths.

 

[Artsfols]

A small transducer is able to reproduce higher frequencies and these are "faster" than lower frequencies - that is all there is to it. Within a given bandwidth, all loudspeakers are equally "fast". There are no "fast" woofers because the "speed" is determined by the highest frequency it reproduces. If it is bass managed, that is typically 80 Hz. The room resonances are the "slow" part, and unless these are controlled, bass will be "slow", as MaxwellsEq said. Science is different from subjective reviewing - facts are involved.

I take on board what you are saying although it's not exactly my more primitive language so it took a bit.
So when I am talking about velocity, I'm not talking about the speed of the cone moving through air. I'm talking about the ability of the cone to move from say, producing a "C" note to producing a "D" note, and how long that transition takes in relation to the time it takes in the source. The cone is vibrating at one speed, and then must shift to vibrating at another speed.
To enlarge the question a little bit. (First, I'm not sure I completely understand how a cone can vibrate at 261.6 HZ (middle C) and 293.6 Hz (D) at the same time. My view is that it would produce a more complex wave encapsulating both frequencies, so that if you measured the wave lengths crest to crest, you'd get a variety of distances due to the intermingling of the two notes, and the speaker then produces that rather dissonant wave. Tell me if that is wrong.)
Now consider a full orchestra, with not only all kinds of notes, but various identifying harmonics that create the timbre of each unique instrument.
So what I think you have told me is that if a speaker can produce each tone from say 20 HZ to 20 kHZ, which is well outside my audible range, and also reproduce faithfully the gain level of the source (flat response), then that speaker will have no problem reproducing the fidelity of a full orchestra.
That is problematic for me. Why do I still need 5, 7 or more speakers to reproduce spatiality? I only can hear 2 waves even with 8 speakers, I should be able to reproduce any spatial effect through headphones, i.e. any directionality. Maybe I can ...
Further, even with my more limited hearing range now topping out at 10kHz, plus tinnitus so I miss very low level sounds, a live orchestra still sounds better to me than any recording. It's true I can turn my head to pick out what the harp is playing, say, which I cannot do with my stereo, but the overall fidelity and ability to hear individual instruments simultaneously can not be matched by my reasonably good stereo equipment.
So, I think there is more to the art of reproduction than just a flat frequency response, and I have trouble believing that a flat frequency response is a complete measure of musical fidelity.
Don't get me wrong on this, I am a strong believer in the efficacy of lab tests. But I still need to hear the speaker as well. For electronic equipment, I'll buy without hearing, strictly on the specs.

 

[Artsfols]

You don't hear that "speed." You hear pressure waves. And those waves must follow simply linear relationship between frequency, wavelength and hence, velocity (which is a simple function of the former two).

Think of it this way. Have a driver playing at very faint volume where you can't even see the cone moving. Then crank it up where it is playing at is maximum excursion. You think the sound changes because the cone has farther to travel in the second case?

I would like to fully understand what you're saying, even though it's purely academic.
At the lower volume the cone is moving very little, but let's say vibrating at 440 HZ to produce the note A.
If we turn up the sound, I thought the cone would still vibrate at 440 HZ but would excurse a greater distance. Thus the cone is excursing at a higher velocity, is it not? The wave coming out is at the same velocity at both volumes, but the cone surface velocity has increased.
(This has nothing to do with my original question around speaker transitions, but now that we're on this path.)

And while we're on this, it comes to mind the objection to electrostatic speakers, considered the ultimate in sound fidelity. But the issue with them is one of dynamics. They have trouble with volume shifts. This observation comes from reading reviews, not personal experience, so it is hearsay. But it makes sense to me. Going back to my C and D note. If a very loud C is played together with a soft D, and then suddenly the D is hammered, and the C fades away. Is there a limit to how well a speaker can reproduce complex dynamics. It not only has to shift from soft to loud, but simultaneously shift from loud to soft on a different note.

 

[Floyd Toole]

I take on board what you are saying although it's not exactly my more primitive language so it took a bit.
So when I am talking about velocity, I'm not talking about the speed of the cone moving through air. I'm talking about the ability of the cone to move from say, producing a "C" note to producing a "D" note, and how long that transition takes in relation to the time it takes in the source. The cone is vibrating at one speed, and then must shift to vibrating at another speed.
To enlarge the question a little bit. (First, I'm not sure I completely understand how a cone can vibrate at 261.6 HZ (middle C) and 293.6 Hz (D) at the same time. My view is that it would produce a more complex wave encapsulating both frequencies, so that if you measured the wave lengths crest to crest, you'd get a variety of distances due to the intermingling of the two notes, and the speaker then produces that rather dissonant wave. Tell me if that is wrong.)
Now consider a full orchestra, with not only all kinds of notes, but various identifying harmonics that create the timbre of each unique instrument.
So what I think you have told me is that if a speaker can produce each tone from say 20 HZ to 20 kHZ, which is well outside my audible range, and also reproduce faithfully the gain level of the source (flat response), then that speaker will have no problem reproducing the fidelity of a full orchestra.
That is problematic for me. Why do I still need 5, 7 or more speakers to reproduce spatiality? I only can hear 2 waves even with 8 speakers, I should be able to reproduce any spatial effect through headphones, i.e. any directionality. Maybe I can ...
Further, even with my more limited hearing range now topping out at 10kHz, plus tinnitus so I miss very low level sounds, a live orchestra still sounds better to me than any recording. It's true I can turn my head to pick out what the harp is playing, say, which I cannot do with my stereo, but the overall fidelity and ability to hear individual instruments simultaneously can not be matched by my reasonably good stereo equipment.
So, I think there is more to the art of reproduction than just a flat frequency response, and I have trouble believing that a flat frequency response is a complete measure of musical fidelity.
Don't get me wrong on this, I am a strong believer in the efficacy of lab tests. But I still need to hear the speaker as well. For electronic equipment, I'll buy without hearing, strictly on the specs.

You have understood the first part - that having a flat frequency response and adequate bandwidth are basic necessities to reproducing the combined pressure waveforms created when sounds from multiple instruments combine. All the information is in the pressure waveform, as our own ears tell us. This is still a bit simplified, because the rule is true only for minimum-phase technical devices: microphones, loudspeaker transducers, and amplifiers. Phase shift can distort the waveforms, but it turns out that humans don't hear it, especially when listening in normal rooms. We don't hear "accurate" waveforms in reflective spaces - like normal listening rooms or concert halls.

Interesting that you would buy electronics based on - frequently biased - specifications, but not loudspeakers. We now have meaningful measurements, as Amir, Erin, and others now publish to the great benefit of all serious audiophiles. I would trust that data more than my own ears in an uncontrolled listening situation.

What you next need to appreciate is how two ears and a brain allow us to perceive direction and space. Yes, you "only can hear 2 waves even with 8 speakers" but the human binaural hearing system allows you to appreciate that there are 8 speakers in the room. Stereo it isn't. Stereo is a basic problem for our industry, and it has become the default format. I dig into the details of this in the upcoming 4th edition of my book and the insights are very interesting. The phantom images that populate the soundstage between the loudspeakers are not comprised of accurate spectra or waveforms - both loudspeakers "talk" to both ears and there is comb filtering, especially noticeable for the featured artist in the centre location. A problem with multichannel audio is that the centre loudspeaker - a good idea - sounds different from the other phantom images on the soundstage. It is a challenge for recording engineers to deal with the centre channel. Humans are very adaptive - and forgiving! However, good immersive multichannel recordings can be remarkably impressive, allowing one to walk around the room and not lose the illusion.

Headphones, and the related cross-talk cancelled loudspeaker version, are fundamentally different. This is where 2 ears and 2 channels make sense, but recordings are mixed for loudspeaker stereo and what is heard in headphones is not what was intended. With technically accurate headphones and well synchronized head tracking binaural (dummy head) recordings through headphones can be remarkably like "being there". These technologies are discussed in the upcoming book, and they are serous options.

 

[Artsfols]

You have understood the first part - that having a flat frequency response and adequate bandwidth are basic necessities to reproducing the combined pressure waveforms created when sounds from multiple instruments combine. All the information is in the pressure waveform, as our own ears tell us. This is still a bit simplified, because the rule is true only for minimum-phase technical devices: microphones, loudspeaker transducers, and amplifiers. Phase shift can distort the waveforms, but it turns out that humans don't hear it, especially when listening in normal rooms. We don't hear "accurate" waveforms in reflective spaces - like normal listening rooms or concert halls.

Interesting that you would buy electronics based on - frequently biased - specifications, but not loudspeakers. We now have meaningful measurements, as Amir, Erin, and others now publish to the great benefit of all serious audiophiles. I would trust that data more than my own ears in an uncontrolled listening situation.

What you next need to appreciate is how two ears and a brain allow us to perceive direction and space. Yes, you "only can hear 2 waves even with 8 speakers" but the human binaural hearing system allows you to appreciate that there are 8 speakers in the room. Stereo it isn't. Stereo is a basic problem for our industry, and it has become the default format. I dig into the details of this in the upcoming 4th edition of my book and the insights are very interesting. The phantom images that populate the soundstage between the loudspeakers are not comprised of accurate spectra or waveforms - both loudspeakers "talk" to both ears and there is comb filtering, especially noticeable for the featured artist in the centre location. A problem with multichannel audio is that the centre loudspeaker - a good idea - sounds different from the other phantom images on the soundstage. It is a challenge for recording engineers to deal with the centre channel. Humans are very adaptive - and forgiving! However, good immersive multichannel recordings can be remarkably impressive, allowing one to walk around the room and not lose the illusion.

Headphones, and the related cross-talk cancelled loudspeaker version, are fundamentally different. This is where 2 ears and 2 channels make sense, but recordings are mixed for loudspeaker stereo and what is heard in headphones is not what was intended. With technically accurate headphones and well synchronized head tracking binaural (dummy head) recordings through headphones can be remarkably like "being there". These technologies are discussed in the upcoming book, and they are serous options.

You may note that I specifically mentioned headphones in conjunction with reproducing directionality. I found your comment on that interesting. There's no question that stereo speakers could not reproduce an arbitrary directionality; they are directional in their own way.
The reason I trust specs on electronics is that they are relatively neutral devices. What goes in is what goes out, and if it doesn't the tests will pick that up. I do not trust that speakers (or microphones) measurements tell the whole story. I could be convinced that they do, but at this point I'm not there.

 

[Floyd Toole]

There's no question that stereo speakers could not reproduce an arbitrary directionality; they are directional in their own way.

However, two loudspeakers in a cross-talk cancellation configuration, in a room with some reflection control, can do a credible job of delivering binaural recordings - convincingly externalized - and with synthesized phantom loudspeakers at +/- 3- deg. better stereo than I have ever heard in conventional setups. The walls and loudspeakers essentially disappear. That is why I said they are serious options.

Historically, loudspeaker measurements have been relatively uninformative, and specifications useless. However that has changed, and I had something to do with it. I guess you are not aware of my research, and the fact that is is the basis for the Spinorama format of loudspeaker measurements and data presentation. There is massive subjective/objective data supporting it, not my opinion.

 

[Artsfols]

However, two loudspeakers in a cross-talk cancellation configuration, in a room with some reflection control, can do a credible job of delivering binaural recordings - convincingly externalized - and with synthesized phantom loudspeakers at +/- 3- deg. better stereo than I have ever heard in conventional setups. The walls and loudspeakers essentially disappear. That is why I said they are serious options.

Historically, loudspeaker measurements have been relatively uninformative, and specifications useless. However that has changed, and I had something to do with it. I guess you are not aware of my research, and the fact that is is the basis for the Spinorama format of loudspeaker measurements and data presentation. There is massive subjective/objective data supporting it, not my opinion.

Don't get me wrong. I don't have a position, but that mainly comes out of a lack of knowledge. I appreciate your efforts to enlighten me. Next I Google "Spinorama" but probably ... tomorrow.

 

[amirm]

To enlarge the question a little bit. (First, I'm not sure I completely understand how a cone can vibrate at 261.6 HZ (middle C) and 293.6 Hz (D) at the same time.

A speaker plays time domain signal. It is not a frequency analyzer. When those two notes were recorded, it is their sum that is in the recording. That sum has the two notes and intermodulation of them. The speaker just gets that varying voltage and (ideally) linearity changes its cone position to it. There could be two note or thousands (as would be the case with real piano notes). Speaker blindly moves to that change in frequency.

Now, if it can't perform ideally, it produces distortions but that doesn't change the fundamental way it works.

 

[Artsfols]

A speaker plays time domain signal. It is not a frequency analyzer. When those two notes were recorded, it is their sum that is in the recording. That sum has the two notes and intermodulation of them. The speaker just gets that varying voltage and (ideally) linearity changes its cone position to it. There could be two note or thousands (as would be the case with real piano notes). Speaker blindly moves to that change in frequency.

Now, if it can't perform ideally, it produces distortions but that doesn't change the fundamental way it works.

"if it can't perform ideally, it produces distortions"
So what I am really trying to get at here ... are speakers strained, or is the reproduction accuracy reduced ... due to characteristics of the source material such as the number of notes, the harmonics that have to be produced, the number of instruments, the duration of notes .. characteristics other than purely volume and dynamics, or the frequency range, which do affect speaker performance.
(I think I've already heard that they are not, but just to be sure.)
I'm also not sure what the sentence "it is not a frequency analyzer" means. What I do know is that the source notes C and D from a piano move air at specific frequencies, and a speaker has to move air at those exact same specific frequencies and with the same relative volume to each other and over time. If it does not then it is deficient in some way.
Below is an output from AI on what the combined waveform of notes C and D looks like, which of course is no mystery to you. I am just trying to show where my thinking is going so you can tell me where I might be wrong.
To me what is interesting is that a single wave can sound to the brain as two notes, not one. Not only can the human brain decode a single wave into two notes, it can decode a single very complex wave into dozens of simultaneous notes, and the timbre that identifies individual instruments and so on.
It strikes me then that a speaker has to reproduce a single complex wave with incredible accuracy, or you lose definition.
My concern with speakers is whether they can follow frequency and volume changes accurately as those waves become more complex. We know the electronics can. But a speaker cone is a physical thing with properties such as stiffness or lack of stiffness, not to mention its mounting, cabinet properties and so on. Then the speaker cone must also respond accurately to changes in frequency, moving from vibration at one frequency to another in a milli-second. I'm not concerned about resonances which are more an issue of cabinet design, but differences in reproduction accuracy due to the nature of the source sound.
In actual experience, orchestral sound can be very muddy on cheap speakers. Instrument timbre especially is reduced.
So the assertion is that lack of speaker performance will reveal itself in a frequency response graph built using individual tones that do not resemble the complexity of the sound found, say, in an orchestral performance. Or maybe they do. I really don't know. I have run room tests using Audyssey in my Denon amp, and basically the amp sends out a few squawks and gets a frequency response graph from them. In what those squawks consist I don't know. Would I get exactly the same frequency response graph if I played an orchestral recording and created a frequency response graph from it - based on anomalies, or difference from source to output, of course.

 

[MattHooper]

You have understood the first part - that having a flat frequency response and adequate bandwidth are basic necessities to reproducing the combined pressure waveforms created when sounds from multiple instruments combine. All the information is in the pressure waveform, as our own ears tell us. This is still a bit simplified, because the rule is true only for minimum-phase technical devices: microphones, loudspeaker transducers, and amplifiers. Phase shift can distort the waveforms, but it turns out that humans don't hear it, especially when listening in normal rooms. We don't hear "accurate" waveforms in reflective spaces - like normal listening rooms or concert halls.

I’m curious about what, if anything at all, frequency response leaves out in telling us how loudspeaker will sound.

As I understand you’ve written that most things that matter will show up in the frequency response - including driver or cabinet colorations due to resonances (please correct me if I’m wrong).

I know there’s of course off axis measurements, and similarly neutral speakers can have different dispersion patterns which can cause them to sound different depending on the room and set up. But that’s still essentially about frequency response.

So my question is, what of importance does frequency response leave out, if anything?

So two scenarios:

1. If you were comparing (blind) loudspeaker A and B, and loudspeaker A seems to be revealing slightly more Sonic detail, for instance very subtle recorded room acoustics or very subtle artificial reverb is revealed on speaker A, would the explanation of that difference likely be spotted in the frequency response of each speaker? Or is it possible for two speakers to measure essentially neutral, but one has a slightly livelier cabinet, or the other design has gone to heroic measures to make a completely inert cabinet. Is it possible that Sonic advantages from a more inert cabinet might reveal the type of subtle details I’m talking about, and this factor would not show up in or be obvious from evaluating just the frequency response of the two speakers? Or will any such audible difference at all in the designs be captured in the frequency response?

I guess another way of putting it: as long as you have a flat frequency response, will that assure the speaker is producing all the recorded detail un-obscured, and no other parameter of the design would affect this?

2. There’s been occasional discussions about the perceptual effects of larger speakers and larger drivers versus smaller speakers and smaller drivers. The idea of being that there is something about a truly big speaker that can give a sense of larger scale versus a smaller speaker, even if they both have the same frequency response specifications.

So as a thought experiment: let’s say we had A super Duper Speaker comparator system for blind testing they could accommodate any size speaker. And let’s say we have some “ pick your neutral” stand mounted monitor (say a good Genelec monitor) combined with a subwoofer or two, and we’ve got those measuring with flat frequency response from 20 Hz to 20 K.

On the other hand, we have a huge PA system, of the type you might find in a very large club, and this system has also been made flat from 20 Hz to 20 K.

So in terms of strict frequency response, they are essentially the same. And we play them back at the same sound levels.

And yet, at least intuitively, I think we’d expect that even playing at something like an 85 DB level, the systems would sound different, the scale of the PA system sounding significantly more grand. And probably more convincing in various ways for instance recreating an orchestra, etc.

Is there anything wrong with these intuitions?

And if not, what would count for the Sonic differences between these two systems that have the same frequency response played at the same volume levels? Is it something to do with the sheer amount of air being moved by the larger system? Or is that not even a factor if we’re talking about the same levels and the same frequency response?

(Thanks to anybody else who wants to chime in on this).

 

[Moonlightshadow]

Well designed larger drivers in their well chosen passband reproduce the same input signals exactly the same well as designed smaller ones, otherwise this would be measured as higher distortion (linear and non-linear) which is the deviation to the input signal. This shows that "fast loudspeakers" is one of the unfortunately widely spread typical audiophile urban myths.

"Faster" loudspeakers just play at a higher frequency

 

[amirm]

I'm also not sure what the sentence "it is not a frequency analyzer" means. What I do know is that the source notes C and D from a piano move air at specific frequencies, and a speaker has to move air at those exact same specific frequencies and with the same relative volume to each other and over time.

That's not correct way to look at it. If you are playing a cord on the Piano, all the notes combine into a composite waveform that the microphone picks up to record. Neither the microphone, nor the recorder know anything about its frequency make up. As long as they both have ample bandwidth, they record the totality of what is in that cord. Same happens at playback. The whole system is just varying voltages.

Now, when you hear it, or when we want to analyze the sound, we decompose the time domain, voltage varying signal into frequency components. But neither microphone, nor speaker are operating that way. If the signal is a mix of low and high frequencies, the high frequencies show up as small variations on the low frequency waveform (assuming the high frequencies are lower amplitude).

As an example, look at this 32-tone signal in frequency domain:

This is what it looks like to a microphone or speaker:

Speaker and microphone simply move their cones proportional to this latter waveform, oblivious to the fact that it is made up of 32 distinct tones.

 

[amirm]

So what I am really trying to get at here ... are speakers strained, or is the reproduction accuracy reduced ... due to characteristics of the source material such as the number of notes, the harmonics that have to be produced, the number of instruments, the duration of notes .. characteristics other than purely volume and dynamics, or the frequency range, which do affect speaker performance.

The deviations in response of speaker either manifest themselves in linear changes in frequency response, or non-linear changes due to distortion. Neither is characterized by slow or fastness of anything. You can have a lousy 15 inch woofer that produces far more distortion playing 50 Hz than a small one that is more linear. Speed/velocity are not terminology we use here to describe the artifacts. In my tests, I show both frequency response and distortion so what you worry about, is covered in the measurements.

 

[Heinrich]

"if it can't perform ideally, it produces distortions"
...
So the assertion is that lack of speaker performance will reveal itself in a frequency response graph built using individual tones that do not resemble the complexity of the sound found, say, in an orchestral performance.

Please do not confuse the results for linear characteristics with the measurements aimed at non-linear components. When measuring the linear ones, such as frequency response, the latter are neglected.
Moreover, frequency is a concept that has to do with the human passion for countable phenomena. Frequency is repetition per time, which is relatively robust to measure if the concept is defined mathematically, including its implications. For physical reality, however, the description, which in a certain sense is superimposed, definitely plays no role.
Accordingly, a “rapid change” of frequency from A to B, as you describe, is also not a physically meaningful parameter. Among other things, because a loudspeaker simply has no memory; how can it recognize repetition?
The given measurements follow mathematically logical, conceptually sound considerations. This is an advantage, especially if new concepts such as “fast” are to be introduced and these are to be somehow compatible beyond a certain linguistic culture.
In any case, I would be reluctant to forfeit the concept of frequency: a rocket is fast, a loudspeaker membrane is terribly slow.

+/-3mm in a twentieth of a second makes around 18cm/s max, i.e. 0.6km/h, a tenth of my walking speed ... and then it again has no memory for resolving repetition.

 

[Mauro]

You have understood the first part - that having a flat frequency response and adequate bandwidth are basic necessities to reproducing the combined pressure waveforms created when sounds from multiple instruments combine. All the information is in the pressure waveform, as our own ears tell us. This is still a bit simplified, because the rule is true only for minimum-phase technical devices: microphones, loudspeaker transducers, and amplifiers. Phase shift can distort the waveforms, but it turns out that humans don't hear it, especially when listening in normal rooms. We don't hear "accurate" waveforms in reflective spaces - like normal listening rooms or concert halls.

Interesting that you would buy electronics based on - frequently biased - specifications, but not loudspeakers. We now have meaningful measurements, as Amir, Erin, and others now publish to the great benefit of all serious audiophiles. I would trust that data more than my own ears in an uncontrolled listening situation.

What you next need to appreciate is how two ears and a brain allow us to perceive direction and space. Yes, you "only can hear 2 waves even with 8 speakers" but the human binaural hearing system allows you to appreciate that there are 8 speakers in the room. Stereo it isn't. Stereo is a basic problem for our industry, and it has become the default format. I dig into the details of this in the upcoming 4th edition of my book and the insights are very interesting. The phantom images that populate the soundstage between the loudspeakers are not comprised of accurate spectra or waveforms - both loudspeakers "talk" to both ears and there is comb filtering, especially noticeable for the featured artist in the centre location. A problem with multichannel audio is that the centre loudspeaker - a good idea - sounds different from the other phantom images on the soundstage. It is a challenge for recording engineers to deal with the centre channel. Humans are very adaptive - and forgiving! However, good immersive multichannel recordings can be remarkably impressive, allowing one to walk around the room and not lose the illusion.

Headphones, and the related cross-talk cancelled loudspeaker version, are fundamentally different. This is where 2 ears and 2 channels make sense, but recordings are mixed for loudspeaker stereo and what is heard in headphones is not what was intended. With technically accurate headphones and well synchronized head tracking binaural (dummy head) recordings through headphones can be remarkably like "being there". These technologies are discussed in the upcoming book, and they are serous options.

Who is the manufacturer that sells a system / signal processor with “cross-talk cancelled loudspeaker version”?
You got me curious. What should I buy to experience this? I have Revel Performa f206, if it is relevant..

 

[thewas]

"Faster" loudspeakers just play at a higher frequency

That's why I wrote also "in their well chosen passband" where such wouldn't play a role.