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

[MattHooper]

I had a listening experience sometime in the 90's that was something of an epiphany. But in a negative way. I had an extended demo at one of New York's big High End stores of the first "super speaker" I'd ever heard: the giant, legendary IRS V reference system.

I listened to a variety of classical music pieces that I knew well (at that point I regularly attended orchestral performances). It was something of a shocking experience hearing the Vs. The clarity, detail, boxless presentation and especially the SCALE of the sound. Here before me, orchestral music was being recreated with something like the scale, or impression, of the real thing (or at least I'd never heard something so close in scale).

I'd close my eyes and the orchestra would just spread out huge in front of me. The sense of being able to hear each instrument and section, full sized and clearly, was amazing.

But something hit me about the sound, especially when I had my eyes closed asking "could this be the real thing?" The fact that the scale of the sound was so close to the real thing seemed to make what was missing stick out in relief. What was missing was...timbral realism.

Closing my eyes while listening to real symphonies (as I always did), I'd hear an incredible sonic richness and variety in orchestral timbre between sections and individual instruments - the particular brassiness of horns, the particular woody resonance of a string instrument, the particular combination of reed/metal/warmth of the wind instruments - there was a sort of unlimited sense of resolution and tonal/timbral surprise and color.

But the orchestra in front of my closed eyes on the IRS Vs sounded bland, grayed, homogenized, electronic versions. It was almost like a real orchestra where the musicians instruments had been replaced by plastic replicas, simultaneously making no instrument sound exactly as it normally does, and homoginizing the timbre of all the instruments.

It was fairly disconcerting.

I left thinking about an explanation for this and really could only come up with a superficially intuitive answer: Well, now that I think of it, of course asking a speaker system isn't going to be able to fully reproduce all the real-world timbres of so many different materials! Why would we think as system producing sound by vibrating just one or two materials (e.g. woofers/tweeters) could sound just like a thousand different materials? Of course it's going to be homogenizing everything!

As I said, that seemed a superficially intuitive answer. But I don't think it really accounts for the nature of sound and perception. The fact is that even the sound of a whole orchestra arrives as essentially a mashed together wave-form, and it ends up vibrating an eardrum, tiny bones, hair cells etc. We don't need every type of material in our ears in order to "hear" all the materials accurately of an orchestra. If sound perception worked like that, sound recording (microphones) and reproduction as it is wouldn't work at all. Obviously the full sound of all those instruments must, in principle, be able to be represented in aggregate, from microphone to speaker system to our ears. Again: in principle.

So the question is, variously: Why DOESN'T reproduced sound produce the timbre of the real thing with great accuracy. Or...CAN and DOES it?

John Dunlavy used to claim his speakers could reproduce the sounds of real instruments so accurately that in live vs recorded demos people couldn't tell the difference. (Though he used something of a cheat as I remember...a sort of syllogistic line of inference that involved headphones).

So this subject could go in various directions, but I guess my main questions are:

Do you think sound systems can truly, accurately reproduce the sound, and exact timbre, of the real thing? (Voices, acoustic instruments etc). If so, how? If not, why not? And...do you care? Is that something you want out of your system as any sort of goal?

I've seen various takes on this. Many audiophiles are obsessed with recreating realism, having a system that can sound like the real thing - something like The Absolute Sound. This is often poo-pooed by the faction who say "You'll never be able to reproduce that so why try?" And it seems some people settle instead for a different type of goal - accuracy to the electronic properties of the source signal. So long as the signal contained in the recording is reproduced with as little distortion as possible, well...that's the sound you get and "I'm good with that and don't expect anything more." Sometimes a great recording may seem very realistic, a bonus, that that's not the overriding goal.

What are your thoughts?

 

[SIY]

A stereo pair of speakers reproducing a stereo recording will never sound like the real thing. The limitation is... stereo.

 

[MattHooper]

A stereo pair of speakers reproducing a stereo recording will never sound like the real thing. The limitation is... stereo.

Thanks, but that's a bit vague.

Do you mean it will never be able to recreate the real thing spatially? Or timbrally? If not the latter, why not? What about a mono speaker playing a mono recording of an instrument, or instruments?

 

[SIY]

You assume the issues are separable. I do not.

And a single speaker will not in general have the polar pattern of a lone instrument.

 

[RayDunzl]

What was missing was...timbral realism.

[Thing-Fish:]
What on urf do you mean: 'Timbral Realismum'? I got yo language hangin', boy, 'long wif a two-week supply of IGNINT McNUGGET, de breakfast o' champiums!

 

[RayDunzl]

Sorry.

I did find this, though:

https://www.research.ed.ac.uk/portal/files/52789744/Glasgow_1.pdf

 

[oivavoi]

Interesting thread. Two brief points:

1) Perceived timbre always changes because of reflections. I would guess that a major reason for the richer timbre in concert halls etc is because the instruments radiate in other ways than most stereo speakers do, and there is more reflected energy in the hall which reinforces the direct sound, and also adds to the perceived timbre.
2) Transducers are a compromise. In the bass they are much smaller than the wavelengths they are asked to reproduce, and in the treble they are larger. I'm pretty sure this affects timbre etc.

That said, I do feel that the best contemporary loudspeakers are able to get pretty close to the timbre of real instruments, when it comes to on-axis direct sound. The limitations of many recordings, stereo limitations, listening room acoustics and the radiation pattern of loudspeakers is another matter though.

 

[Shadrach]

I had a listening experience sometime in the 90's that was something of an epiphany. But in a negative way. I had an extended demo at one of New York's big High End stores of the first "super speaker" I'd ever heard: the giant, legendary IRS V reference system.

I listened to a variety of classical music pieces that I knew well (at that point I regularly attended orchestral performances). It was something of a shocking experience hearing the Vs. The clarity, detail, boxless presentation and especially the SCALE of the sound. Here before me, orchestral music was being recreated with something like the scale, or impression, of the real thing (or at least I'd never heard something so close in scale).

I'd close my eyes and the orchestra would just spread out huge in front of me. The sense of being able to hear each instrument and section, full sized and clearly, was amazing.

But something hit me about the sound, especially when I had my eyes closed asking "could this be the real thing?" The fact that the scale of the sound was so close to the real thing seemed to make what was missing stick out in relief. What was missing was...timbral realism.

Closing my eyes while listening to real symphonies (as I always did), I'd hear an incredible sonic richness and variety in orchestral timbre between sections and individual instruments - the particular brassiness of horns, the particular woody resonance of a string instrument, the particular combination of reed/metal/warmth of the wind instruments - there was a sort of unlimited sense of resolution and tonal/timbral surprise and color.

But the orchestra in front of my closed eyes on the IRS Vs sounded bland, grayed, homogenized, electronic versions. It was almost like a real orchestra where the musicians instruments had been replaced by plastic replicas, simultaneously making no instrument sound exactly as it normally does, and homoginizing the timbre of all the instruments.

It was fairly disconcerting.

I left thinking about an explanation for this and really could only come up with a superficially intuitive answer: Well, now that I think of it, of course asking a speaker system isn't going to be able to fully reproduce all the real-world timbres of so many different materials! Why would we think as system producing sound by vibrating just one or two materials (e.g. woofers/tweeters) could sound just like a thousand different materials? Of course it's going to be homogenizing everything!

As I said, that seemed a superficially intuitive answer. But I don't think it really accounts for the nature of sound and perception. The fact is that even the sound of a whole orchestra arrives as essentially a mashed together wave-form, and it ends up vibrating an eardrum, tiny bones, hair cells etc. We don't need every type of material in our ears in order to "hear" all the materials accurately of an orchestra. If sound perception worked like that, sound recording (microphones) and reproduction as it is wouldn't work at all. Obviously the full sound of all those instruments must, in principle, be able to be represented in aggregate, from microphone to speaker system to our ears. Again: in principle.

So the question is, variously: Why DOESN'T reproduced sound produce the timbre of the real thing with great accuracy. Or...CAN and DOES it?

John Dunlavy used to claim his speakers could reproduce the sounds of real instruments so accurately that in live vs recorded demos people couldn't tell the difference. (Though he used something of a cheat as I remember...a sort of syllogistic line of inference that involved headphones).

So this subject could go in various directions, but I guess my main questions are:

Do you think sound systems can truly, accurately reproduce the sound, and exact timbre, of the real thing? (Voices, acoustic instruments etc). If so, how? If not, why not? And...do you care? Is that something you want out of your system as any sort of goal?

I've seen various takes on this. Many audiophiles are obsessed with recreating realism, having a system that can sound like the real thing - something like The Absolute Sound. This is often poo-pooed by the faction who say "You'll never be able to reproduce that so why try?" And it seems some people settle instead for a different type of goal - accuracy to the electronic properties of the source signal. So long as the signal contained in the recording is reproduced with as little distortion as possible, well...that's the sound you get and "I'm good with that and don't expect anything more." Sometimes a great recording may seem very realistic, a bonus, that that's not the overriding goal.

What are your thoughts?

No.The moment the pressure waves hits the first transducer the music is gone.
What do you believe?

 

[solderdude]

The recorded sound has individual waves traveling in all directions differently from each instrument.
Microphones loose that directional spread and the angle under which the waves had hit the microphone (depends on the microphone used if it has a certain frequency/angle dependent roll-off). The microphone(s) are just certain points in a 3D plane.
Speakers will just reproduce the summed microphone signals from that point and apply their own directional waves.
It is and can never be a true reproduction as there is no (commercial at least ?) 'holographic' way of recording and reproducing sound.
They can be quite convincing though when the recording itself is.

Some speakers and recordings can sound surprisingly good though. But it is NO 1:1 replacement of the original event that has been recorded.
Surround recordings may get you closer to that experience than stereo will.

 

[BillG]

Do you think sound systems can truly, accurately reproduce the sound, and exact timbre, of the real thing? (Voices, acoustic instruments etc). If so, how? If not, why not? And...do you care? Is that something you want out of your system as any sort of goal?

I'm a melophile, first and foremost, with a science and engineering background. With that in mind, I want my equipment to reproduce the recordings as transparently as possible, within the constraints of the resources I'm willing to allocate to that endeavor - that being a couple of thousand dollars at most.

Do I care if my system sounds exactly like live instrumentation? No, because if I wanted that, I'd invite my many musician friends over to perform. What I do care about is that, I'm enjoying the sounds of the reproduction, and I most certainly do...

 

[Cosmik]

In an audio system, you can think of the entire orchestra being passed through a channel.

If you were to pass a single instrument through that channel, you could probably take some liberties with it and not notice any problems e.g. a non-flat frequency response, resonances, odd phase shifts, distortion. This is because any real instrument or voice is being produced by some highly nonlinear, resonant, mechanical arrangement of vibrating bits and pieces. Adding a bit more of that doesn't stand out.

But if you pass the entire orchestra through a non-neutral channel then the nonlinearities and peculiarities are smeared over the entire orchestra and then they do stand out a mile because they are imbuing all the instruments with this same characteristic.

The only way to hear the timbre of the instruments in the orchestra accurately is to send it through neutral channels. As soon as an audiophile starts to talk about distortion being euphonic, or playing with EQ because they like a bit more 'mid' and so on, they have ceased to think of the orchestra as being separate objects and have begun to think of 'the sound' as a kind of paste whose flavour they are adjusting.

Flavoured paste can be pleasant, but it will never resemble a real orchestra.

Why do audiophiles often eschew orchestral music in favour of solo violin or girl-and-guitar when demonstrating their systems? Because the problems don't stand out as much.

 

[M00ndancer]

As many have stated before, to reproduce a "live" performance you need at least 5.1 surround. Regardless of the performance.

 

[Theo]

Maybe echoic memory is not enough to be sure? Maybe knowing that the sound comes from loudspeakers prevents you from finding the timbre being realistic? (as with sighted test). Anyhow, I'm not sure the timbre is at stake. It's the whole sound field which can't be reproduced accurately (even with a neutral system). You may, as Cosmik pointed out, reproduce a single instrument timbre accurately enough, though.
As to the OP question, I don't necessarily care, sounds good enough to me to enjoy F. Z.

 

[Frank Dernie]

timbral realism.

When I ventured on a 2 year search for my "speakers for a lifetime" about 20 years ago I listened to a lot of fancy speakers. When I went to audition the original Wilson Audio WAMM (the French importer had a used pair for sale) I discovered Goldmund speakers. They were the most accurate in terms of instrumental timbre of any of the highish end speakers I heard during my search.
Given that the timbre of musical instruments is produced by the mix of harmonics added to the fundamental I assume that low harmonic distortion is the most important requirement for timbral accuracy not just even frequency response and whatever directional characteristics are desirable.
One observation that I have made from HiFi News speaker reviews (distortion is one of their measurements) is that quite a few well regarded speakers have poor distortion figures.

 

[andreasmaaan]

Do you think sound systems can truly, accurately reproduce the sound, and exact timbre, of the real thing? (Voices, acoustic instruments etc). If so, how? If not, why not? And...do you care? Is that something you want out of your system as any sort of goal?

There are reasons that this is theoretically impossible, before even going into the technicalities.

Loudspeakers are not fed the "real thing"; they are fed a recording of it (that has probably been mixed and mastered, but that's yet another layer).

Firstly, any recording is taken by a single microphone (or a number of single mics) in an acoustic space that is different from the space in which the loudspeakers are reproducing it.

Our brains use interaural time differences and interaural intensity differences (differences in timing and intensity/amplitude response between sounds arriving at each ear) to localise real acousic events in real spaces. A microphone recording a real sound to be reproduced by a loudspeaker does not record this information, so the loudspeaker therefore cannot reproduce it.

And even if a binaural dummy head mic setup is used to record a real acoustic event (which ensures the ITDs and IIDs are close to correct), a pair of loudspeakers in a room cannot reproduce this to our ears, because the sound from both loudspeakers arrives at both ears of the listener, i.e. there is crosstalk that is not present when we hear a real acoustic event.

In terms of timbre, the polar response of a loudspeaker can never match the polar response of more than one particular real instrument or voice. And even if it did that, its polar response would be incorrect for all other instruments/voices.

Finally, in stereo reproduction, the reflected sounds from the speakers arrive with different delays than the reflections would from a real instrument/voice in the location of the stereo image. So the phase and timing of the reflections must be inherently wrong.

Following from all of the above, the only situation I can conceive of in which a reproduction system could properly mirror the sound of the original acoustic event would be where the acoustic event is recorded and then reproduced by a single high-performance loudspeaker in an anechoic chamber. That would sound awful (anechoic chambers do sound awful), but I believe that loudspeakers are capable of low enough distortion and accurate enough on-axis frequency response these days to make the reproduction indistinguishable from the original, at least for most listeners.

 

[PierreV]

I had the opportunity to compare, in my room, a pro opera singer singing live on an orchestra recording with the same full recording of her + orchestra. In terms of accuracy, naturalness, it was pretty close... once I retreated 10 meters or so. Level matching was obviously extremely hard and, at that range, plus I probably would have blown my speakers or ears if I had tried to match her level

We also listened to a few of her more pedestrian recordings (solo + piano, standard French and German songs). She liked how the piano sounded, but noted that there were differences based on the day the piece was recorded (it was a multi-day recording). She did not remember if things had moved or if the settings had changed, but she drew my attention to those slight changes.

With no direct input, those things would have flown so far above my head and ears that I have abandoned all discrimination hopes. Did the piano sound more "natural" on day 1 or on day 2? How should I know?

 

[daftcombo]

It is like asking why a photo or a movie don't seem as real as real life?

 

[FrantzM]

It is like asking why a photo or a movie don't seem as real as real life.

THANK YOU!!!

 

[DonH56]

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

No, not really.

No.

 

[daftcombo]

But you can have BETTER sound than in real-life, even more REALISTIC, with a dCS Debussy.


More seriously, has any attempt of an ABX between real musicians VS Hi-FI system ever been made?