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

[Ilkless]

https://fmq.fi/articles/loudspeaker-orchestra

This topic made me think of this interesting experiment using anechoically-recorded instruments to create a virtual orchestra when placed in a hall. I first heard of it from a presentation by David Griesinger.

 

[MattHooper]

So the problem seems to be that there is some disagreement about the degree to which a loudspeaker can reproduce truly accurate instrumental timbres. Some seem to say "impossible, due to factors that start at the recording chain and end with problems at the loudspeaker - so don't bother with that as a goal - just stick to whether it's reproducing the signal in a way the original engineers heard it. "

Others seem to think "it can get close, so even if not totally accurate, it's still a reasonable goal to ask a speaker to reproduce naturally recorded instruments and voices with some accuracy and realism."

And, everything in between.

And unfortunately there doesn't seem to be much direct scientific experiment in that direction to settle things: very little if any strictly controlled "live vs reproduced" blind comparisons. And the most robust scientific research we have in terms of blind speaker tests - e.g. NRC/Toole/Harman Kardon et al - is more focused on discerning what people find "pleasing" in a loudspeaker, vs whether a loudspeaker is sounding "accurate and real to the timbre/sound of real life sounds."

That seems to leave folks like me who really value a certain sense of timbre/tone in a speaker, especially where it seems a speaker is reproducing something "right" about the timbre of an instrument, in something of a lurch. I don't see that I can look to any particular set of measurements or research to guide me to a speaker that will, to my perception, "remind me more often of the real thing."

After all, as I've mentioned, I've heard speakers like the Revels which are great, but I've had the sensation of "hearing something more like the real thing" from other speakers that don't measure like the Revels.

On that note, I like to switch around my speakers - I went from listening to my Thiel 2.7s, to my Waveform monitors (very neutral, excellent dispersion characteristics). I've also spent quite a bit of time listening to Vivid Audio speakers, including lots of stop overs at my Pal's place when he had the Vivid Audio Kaya 45 speakers. The Vivid speakers tend to measure quite well in the Soundstage NRC measurements. They sounded super clean, clear, neutral etc.

But as I've mentioned, I find that every speaker I've ever heard tends to have a "voice" - a timbre if you will - which is stamped on everything.
A homogenizing effect. Which reminds me "these sounds are being produced by speaker drivers, not the real thing." John Atkinson made some insightful comments in some of his speaker reviews especially about drum cymbals, praising a speaker that made drum cymbals, especially high-hats - sound properly like sticks hitting metal, more like the real thing, rather than "bursts of modulated white noise." That "modulated white noise" effect is something I'm sensitive to and I immediately notice that really listening carefully to drum cymbals on many speakers doesn't survive that scrutiny. It just sounds like tweeters and midrange drivers modulating noise to mimic cymbals, but not *really* sounding distinctive like the real thing. And that particular observation is a symptom of the general sense of homogenization that I seem to hear among different speaker systems.

That's one reason why it really spins my head around when I hear a speaker *actually* produce a compellingly wide variety of instrument timbres.
Which again happened last night when I switched in my MBL radialstrahler 121 omni speakers. As good as the Thiels are, as neutral as the Waveforms and the Vivid speakers are...WOW...sounds through the MBLs just took on a rainbow-like sense of timbral variation. Within a single recording, a wood block sounded SO much like "wood" being struck, cymbals didn't sound like white noise bursts at all, but distinctly "metal being struck in the air" compared to the "wood being struck" near by it. Brass instruments had such a distinctly "brassy object vibrating" distinction from a woodwind's resonating wood body, or the particular timbre of a clarinet's vibrating column. Everything just sounded that much more real.

And I have no idea why this would be, or what would predict this. After all, I've seen even the MBL speakers somewhat dissed by engineer types as being "fun...but not particularly accurate." From which I would have inferred I'd have no reason to expect the type of experience I have when listening to them. What explains this experience they create for me of hearing a wider pallette of timbres? Dunno. This is one reason why it seems I can't be led to what I like in a speaker strictly by measurements, and that ultimately I have to hear a speaker for myself to pick up nuances that I care about, and which to me are more believable and natural sounding. (BTW, another brand of speakers that I find excel in the "producing a wider range of convincing timbres" are the Joseph Audio speakers - another brand I'm considering purchasing at some point).

 

[MRC01]

...
I don't see that I can look to any particular set of measurements or research to guide me to a speaker that will, to my perception, "remind me more often of the real thing."

I think flat (deviations <= 2 dB) frequency response and low distortion are necessary conditions. Whether they are sufficient, I don't know. At least you know, if you don't have at least this, walk away.

...
... I find that every speaker I've ever heard tends to have a "voice" - a timbre if you will - which is stamped on everything.
A homogenizing effect. Which reminds me "these sounds are being produced by speaker drivers, not the real thing." John Atkinson made some insightful comments in some of his speaker reviews especially about drum cymbals, praising a speaker that made drum cymbals, especially high-hats - sound properly like sticks hitting metal, more like the real thing, rather than "bursts of modulated white noise."
...

This is something I notice with planars: they induce less of their own sound, the voicing is more transparent and faithful to whatever the source sounds like. Even they are not perfect, but among the best I've heard. This is not exclusive. I've heard some conventional dynamic speakers having good transparency, it's just less common.

...
That's one reason why it really spins my head around when I hear a speaker *actually* produce a compellingly wide variety of instrument timbres.
...

Transparent playback is necessary but not sufficient. This requires a combination of transparent recordings too. And most recordings are not transparent. Classical music is perhaps closest, the genre least likely to have engineers intentionally adding artificial EQ and compression. But they still do, just to a lesser extent. It's not entirely the engineer's fault: no microphone is perfect, they all have their audible limitations. Choosing the right mic and setup is both art & science. And sometimes, 2 wrongs make a right: the flaws of the recording counteract the flaws in playback creating something that sounds transparent & natural.

Over the years one of the transparency clues I've learned: how different do recordings sound? The more transparent a playback system is, the more different various recordings sound. It steps out of the way revealing the differences.

 

[Blumlein 88]

So the problem seems to be that there is some disagreement about the degree to which a loudspeaker can reproduce truly accurate instrumental timbres. Some seem to say "impossible, due to factors that start at the recording chain and end with problems at the loudspeaker - so don't bother with that as a goal - just stick to whether it's reproducing the signal in a way the original engineers heard it. "

Others seem to think "it can get close, so even if not totally accurate, it's still a reasonable goal to ask a speaker to reproduce naturally recorded instruments and voices with some accuracy and realism."

And, everything in between.

And unfortunately there doesn't seem to be much direct scientific experiment in that direction to settle things: very little if any strictly controlled "live vs reproduced" blind comparisons. And the most robust scientific research we have in terms of blind speaker tests - e.g. NRC/Toole/Harman Kardon et al - is more focused on discerning what people find "pleasing" in a loudspeaker, vs whether a loudspeaker is sounding "accurate and real to the timbre/sound of real life sounds."

That seems to leave folks like me who really value a certain sense of timbre/tone in a speaker, especially where it seems a speaker is reproducing something "right" about the timbre of an instrument, in something of a lurch. I don't see that I can look to any particular set of measurements or research to guide me to a speaker that will, to my perception, "remind me more often of the real thing."

After all, as I've mentioned, I've heard speakers like the Revels which are great, but I've had the sensation of "hearing something more like the real thing" from other speakers that don't measure like the Revels.

On that note, I like to switch around my speakers - I went from listening to my Thiel 2.7s, to my Waveform monitors (very neutral, excellent dispersion characteristics). I've also spent quite a bit of time listening to Vivid Audio speakers, including lots of stop overs at my Pal's place when he had the Vivid Audio Kaya 45 speakers. The Vivid speakers tend to measure quite well in the Soundstage NRC measurements. They sounded super clean, clear, neutral etc.

But as I've mentioned, I find that every speaker I've ever heard tends to have a "voice" - a timbre if you will - which is stamped on everything.
A homogenizing effect. Which reminds me "these sounds are being produced by speaker drivers, not the real thing." John Atkinson made some insightful comments in some of his speaker reviews especially about drum cymbals, praising a speaker that made drum cymbals, especially high-hats - sound properly like sticks hitting metal, more like the real thing, rather than "bursts of modulated white noise." That "modulated white noise" effect is something I'm sensitive to and I immediately notice that really listening carefully to drum cymbals on many speakers doesn't survive that scrutiny. It just sounds like tweeters and midrange drivers modulating noise to mimic cymbals, but not *really* sounding distinctive like the real thing. And that particular observation is a symptom of the general sense of homogenization that I seem to hear among different speaker systems.

That's one reason why it really spins my head around when I hear a speaker *actually* produce a compellingly wide variety of instrument timbres.
Which again happened last night when I switched in my MBL radialstrahler 121 omni speakers. As good as the Thiels are, as neutral as the Waveforms and the Vivid speakers are...WOW...sounds through the MBLs just took on a rainbow-like sense of timbral variation. Within a single recording, a wood block sounded SO much like "wood" being struck, cymbals didn't sound like white noise bursts at all, but distinctly "metal being struck in the air" compared to the "wood being struck" near by it. Brass instruments had such a distinctly "brassy object vibrating" distinction from a woodwind's resonating wood body, or the particular timbre of a clarinet's vibrating column. Everything just sounded that much more real.

And I have no idea why this would be, or what would predict this. After all, I've seen even the MBL speakers somewhat dissed by engineer types as being "fun...but not particularly accurate." From which I would have inferred I'd have no reason to expect the type of experience I have when listening to them. What explains this experience they create for me of hearing a wider pallette of timbres? Dunno. This is one reason why it seems I can't be led to what I like in a speaker strictly by measurements, and that ultimately I have to hear a speaker for myself to pick up nuances that I care about, and which to me are more believable and natural sounding. (BTW, another brand of speakers that I find excel in the "producing a wider range of convincing timbres" are the Joseph Audio speakers - another brand I'm considering purchasing at some point).

You are experiencing what Toole calls the circle of confusion. You criticize those findings saying they focus on finding what people find pleasing vs what is accurate and real. So how do you focus instead on what is accurate and real? The lack of a usable reference is where the confusion starts. You are probably thinking, "easy, I'll record some musical instruments playing and let people compare that to hearing the musicians directly". That is actually very hard to do effectively. And if the result is judged as poor, is it the speaker, the microphone, the aspects of how it is setup etc etc.

What was maybe surprising about what the NRC, Toole and Harman have done is that what people find pleasing also happens to be just what an engineer would have expected. An engineer would have sought to create a speaker with a very flat anechoic response, with low distortion. That is what people appear to prefer. The main contribution from such research is what works best in directionality and off axis colorations. It puts numbers to how much is a detraction, and what kind of properties are positively related.

Maybe I'm forgetting other research that directed them. Maybe @Floyd Toole can fill us in on this.

Now having said that, I too prefer ESL's. They seem to do timbre and transparency better. Or maybe their mis-behaviors amplify that in a way that is detrimental to other aspects of sound. I've an experience common to many, which is hearing live a finger plucked guitar or mandolin. Most speakers seem to miss those initial sounds of the strings. It is like they were fraction of a second late moving and too late for the whole sound. ESLs on the other hand seem to get that almost right as rain. Not quite live, but far closer. They sound 'fast'. Of course we know they aren't necessarily fast as that is a frequency response thing. Also that speakers a little weak in the low end thru midrange can emphasize such sounds to seem faster. But doing EQ to a cone speaker making it more like an ESL doesn't seem to get us there either. It does move it in that direction.

I must say, having had some Harman designed speakers, while I might find this or that thing that another speaker does better, their designs are one of those if you'll just sit back and listen they don't intrude upon music. You can forget about speakers more than most. They get much of it very right.

 

[MattHooper]

You are experiencing what Toole calls the circle of confusion. You criticize those findings saying they focus on finding what people find pleasing vs what is accurate and real. So how do you focus instead on what is accurate and real? The lack of a usable reference is where the confusion starts. You are probably thinking, "easy, I'll record some musical instruments playing and let people compare that to hearing the musicians directly". That is actually very hard to do effectively. And if the result is judged as poor, is it the speaker, the microphone, the aspects of how it is setup etc etc.

Exactly!

Well, not exactly...some caveats. I wouldn't say I 'criticize" the findings because that sounds more like I'm point a finger of blame. Rather I'm just point out what anyone here knows: they aren't meant to discern what I'm wishing to discern. It's not a critique that "realism" in the sense of "accurate to the sound of real acoustic voices and instruments" hasn't been the object of much useful scientific research. Just a fact one has to consider in trying to wade through the problem. There isn't much if any research directly addressing the concern I raise. And, no I don't think it's that easy, far from it. Otherwise I think many more people would have done it, which is why the research isn't there to "help" decide the issue.

As I mentioned earlier, I have indeed recorded familiar instruments and voices to see how different speakers compare to "the real thing" (which sometimes included directly comparing the recording through a speaker to the same instrument or voice). There are obvious variables and problems in there in terms of meeting any rigorous scientific-level of scrutiny. But...nonetheless...the more modest goal didn't seem entirely out of reach. Some speakers DID seem to sound more like the real thing than others.

It does seem in principle to be a more realistic goal, to try to replicate the signal that the engineer heard/created with as little distortion as possible. If indeed everyone at every part of the chain got on board with standards that reduced the "circle of confusion." But...in one sense that goal is also forever out of reach, since the vast majority of recordings we listen to have been produced using a wide variety of monitorrs, and rooms, in which the engineer was hearing specific colorations that are not going to be reproduced if our system is "flat/neutral."

So, one picks one's unreachable goals

I generally agree about the perception of planars, like electrostatics, at least giving the sense of "sounding uncolored" and changing notably the character from recording to different recording. (I owned Quad 63s for quite a while, but still abandoned them because they didn't energize the room in a way that I wanted). I also generally agree about the Harmon speakers: they seem to do "less wrong and more right" than the vast majority of speakers I've heard. The science and engineering certainly counts for something.

 

[MattHooper]

I think flat (deviations <= 2 dB) frequency response and low distortion are necessary conditions. Whether they are sufficient, I don't know. At least you know, if you don't have at least this, walk away.

Yes, that is the intuition I immediately have myself. The more neutral the frequency response, and in general the less distortion the speaker introduces, the more varied any recording should sound....which would also seem to entail such a speaker would be the most likely to approach reproducing the sense of "the real thing" with recordings that get us as close as currently possible.

But...without the ability to do true double-blind live vs reproduced tests we are stuck with some subjectivism here, our own perceptions. Since no recording/loudspeaker combo can truly produce purely accurate timbre, since there is always a compromise there, then we are left to our own devices to decide "this one sounds more real than that one, to me." And since this is picking among compromises, we can come to different conclusions. Depends what aspect of "real life" a speaker may be doing more convincingly than another. Two people hearing a horn system vs an electrostatic speaker may come to entirely different conclusions about which they find to be "most realistic or believable." To the person who notices the boxless, transparent, finely detailed, precise sound of the electrostatic as prominent, THAT may be the speaker he deems most realistic.
The other person may be more inclined to cue in on a sense of force, palpability, dynamic range, air-moving ability etc, and feel the horn system "sounds more real."

 

[MRC01]

What exactly is "real"? These same variations also happen with unamplified acoustic music. I own several flutes that all sound different, and each instrument sounds different when playing in different rooms, or sitting at a different location in the group. Our cello player once tried a different kind of strings that made him sound more like a recording: a bit more projection (louder), but with a bit of extra HF "zing" added to the timbre.
Have you listened to "Drums and Bells" by Chris Dutz? It's a phenomenal recording having incredibly natural lifelike voicing. It's the most natural realistic drum & percussion I have ever heard in an audio system. I'm curious whether others agree it is lifelike, or perhaps my opinion is just an artifact of my own subjective perception and quirks of my audio system.

 

[Cosmik]

All 'sighted' of course. I'm not against that at all, but some 'hypothesis' would be good. *Why* is the electrostatic good? The superior phase response or the directionality at high frequencies, etc.? Or could it be psychology; that a visibly-exotic structure like an ELS is bound to inspire a more positive response than a common or garden rectangular box?

I still notice that audio discussions are a paradox: absolute certainty about how audio works (demanding ordinary measurements because they tell us everything) coupled with complete mystification (the same people describing how 60 year old speakers are the best they've ever heard at reproducing a cymbal but suggesting no reason why that should be).

 

[Blumlein 88]

Exactly!

Well, not exactly...some caveats. I wouldn't say I 'criticize" the findings because that sounds more like I'm point a finger of blame. Rather I'm just point out what anyone here knows: they aren't meant to discern what I'm wishing to discern. It's not a critique that "realism" in the sense of "accurate to the sound of real acoustic voices and instruments" hasn't been the object of much useful scientific research. Just a fact one has to consider in trying to wade through the problem. There isn't much if any research directly addressing the concern I raise. And, no I don't think it's that easy, far from it. Otherwise I think many more people would have done it, which is why the research isn't there to "help" decide the issue.

As I mentioned earlier, I have indeed recorded familiar instruments and voices to see how different speakers compare to "the real thing" (which sometimes included directly comparing the recording through a speaker to the same instrument or voice). There are obvious variables and problems in there in terms of meeting any rigorous scientific-level of scrutiny. But...nonetheless...the more modest goal didn't seem entirely out of reach. Some speakers DID seem to sound more like the real thing than others.

It does seem in principle to be a more realistic goal, to try to replicate the signal that the engineer heard/created with as little distortion as possible. If indeed everyone at every part of the chain got on board with standards that reduced the "circle of confusion." But...in one sense that goal is also forever out of reach, since the vast majority of recordings we listen to have been produced using a wide variety of monitorrs, and rooms, in which the engineer was hearing specific colorations that are not going to be reproduced if our system is "flat/neutral."

So, one picks one's unreachable goals

I generally agree about the perception of planars, like electrostatics, at least giving the sense of "sounding uncolored" and changing notably the character from recording to different recording. (I owned Quad 63s for quite a while, but still abandoned them because they didn't energize the room in a way that I wanted). I also generally agree about the Harmon speakers: they seem to do "less wrong and more right" than the vast majority of speakers I've heard. The science and engineering certainly counts for something.

I can supply the ESL emulation EQ curve for Audacity. I spent nearly 60 seconds to construct it. And nearly 10 minutes of exhaustive listening research to confirm its efficacy. Oh and it is the result of my 35 years experience as an audiophile. With 30 of that being an ESL owner.

I agree simply the exotic appearance and novelty of the big panels and knowing how they operate will impress one to expect something very different. And very different is what you get.

I suppose the best approach to compare live vs speaker/recording would be to have only two musicians. One will take the position of the speaker for each channel. Play and record them ahead of time. Then you can hear them live and hear a speaker with a curtain between the two. See how speakers match up. Or maybe like Harman finds, mono is better. One musician and one speaker. Harman could even record the musician, and replace one of their speakers with a seated musician for such a test. You get to compare three speakers and the real thing.

 

[Floyd Toole]

Maybe I'm forgetting other research that directed them. Maybe @Floyd Toole can fill us in on this.

The details are in my book, so this is a simplification. People talk about "accuracy", "realism", etc. as if humans have perfect auditory memories for something they almost certainly never heard at the time of recording. They certainly never heard the sounds picked up by the microphones placed where listeners' ears never go (a few inches from lips and guitar strings, for example, and above an orchestra). The sound field that is captured and recorded is very different from that one experiences in a live performance, and all of the recorded tracks are mixed, equalized and spatially processed based on what is heard through a pair of non-standardized monitor loudspeakers. The contributions to "realism" by skilled recording engineers are profound. The scarcity of truly impressive recordings is evidence of the substantial chance factor involved in the circle of confusion.

Yet, when exposed to different versions of recordings through loudspeakers listeners with normal hearing exhibit remarkable agreement about which ones are awarded the highest scores. Greybeard audiophiles may or may not agree, especially if, like me, they have been exposed to some excessively loud sounds.

Looking at the comments on response sheets answers the question of what is being judged - it is flaws. Listeners comment extravagantly on things they don't like, often in colorful language, but highly rated speakers get much simpler, flattering descriptions. One can conclude with justification that the "best" loudspeaker is the "least flawed" one.The dominant flaw is resonances - colorations added to all recordings, which are easily identified in the multiple (3 or 4 at a time) double-blind comparison tests we perform. Having multiple speakers allows listeners to perceptually "stream" the timbral contribution of the room (a constant factor) from the timbral contributions of the program (another constant factor), and both from the varying timbral contributions of the different loudspeakers. The experimental method matters! In the end, loudspeakers with the least audible contributions - the most neutral ones - get the highest scores and these have consistently been the ones with the fewest measurable flaws. Flattish and smooth frequency responses indicate an absence of resonances. It is not mysterious.

So, in a real sense,the secret to unravelling the puzzle is how the subjective evaluations are done. Allowing any non-auditory influences to interfere is obviously to be discouraged, but this almost universally happens outside controlled laboratory listening tests. It happens in almost all personally conducted tests and published reviews. Humans are very susceptible, but we hate it when reminded of our capriciousness. I have experienced many instances of listeners disliking the loudspeakers they have brought in for evaluation - shock and dismay. We adapt, and can come to accept many variations of "truth".

Finally, to expect anything resembling perfection from a two channel record/reproduction system is a fantasy, but humans - bless us - are excellent at taking small hints and elaborating them into much larger subjective experiences. Just as well, because we are stuck with stereo for the foreseeable future. That explains some of the differences of opinion about loudspeakers with different directional properties, generating different sound fields in the room. For different listeners, with different hearing performance and expectations, certain versions of the recorded soundtracks are more pleasing. Two sound sources are simply not enough.

 

[RayDunzl]

Or could it be psychology; that a visibly-exotic structure like an ELS is bound to inspire a more positive response than a common or garden rectangular box?

Here's what I see when I look, whether I'm listening to the ESL or the JBL.

Ugly.

Casually listening, I can't tell them apart, reliably, unless in the sweet spot. Different imaging.

Critically listening, at volume, these ESLs stomp those JBLs.

(they are more highly powered, so there's that disadvantage, but it persists at equal lesser volume levels)

 

[MattHooper]

What exactly is "real"? These same variations also happen with unamplified acoustic music. I own several flutes that all sound different, and each instrument sounds different when playing in different rooms, or sitting at a different location in the group. Our cello player once tried a different kind of strings that made him sound more like a recording: a bit more projection (louder), but with a bit of extra HF "zing" added to the timbre.
Have you listened to "Drums and Bells" by Chris Dutz? It's a phenomenal recording having incredibly natural lifelike voicing. It's the most natural realistic drum & percussion I have ever heard in an audio system. I'm curious whether others agree it is lifelike, or perhaps my opinion is just an artifact of my own subjective perception and quirks of my audio system.

For sure, there's a "circle of confusion" even extending in to real sounds for those reasons.

But it seems to me that doesn't disbar all possible assessment of what sounds "natural" or "more real."

If we had a naturally recorded voice and played it through, say, the Revel Salon 2s vs through the speakers on an iphone, surely pretty much everyone will recognize and agree the voice through the Revels sound "more like a real-life voice" even if not perfect. Our auditory memory is unreliable...but not wholly unreliable. It gets more unreliable as you scale down the differing sonic characteristics to ever more similar...but becomes more reliable as you scale those differences upward. In a blind test, I'm not going to mistake Godzilla's famous roar with Kermit The Frog. I can recognize my mother's voice on the phone, or any number of sonic environments. So while it is true and something of a justifiable shibboleth in objective circles "our auditory memory is unreliable" it seems to me we still need to maintain a sense of proportion when making that claim. (And where it stops getting from "pretty reliable" to "simply can't trust it at all" I'm no position to say).

But, again, it doesn't seem like our auditory memory is so bad as to disbar any number of cases where we'd agree "X sounds more real than Y."

As to "what exactly is real?" I'm no arbiter of reality so I can only go by how my own experience of constantly comparing live to reproduced has produced certain perceptions or beliefs. When I hear various instruments being played and I close my eyes, the characteristics that stand our is how timbrally varied and specific it sounds - that instrument made of wood just *sounds* like wood, the drum cymbals sound of real metal, the acoustic guitar strings sound metallic, wholly different from the material the drummer is striking on his. snare, etc. That is the impression that live sound sources leave with me every single time I close my eyes and examine it.

When I close my eyes and examine the sound coming from a speaker, it's not that I know exactly what any particular instrument sounded exactly like in front of the microphones. But I DO know, I believe, that they are playing instruments I'm generally familiar with hearing live and that the live instruments would have a distinctiveness that I'm not hearing through the speaker. So it's not so much "I know exactly what those particular instruments in the musician's hands sounded like in real life" but more "I have a good idea by now how instruments tend to sound in real life."
And some speakers tend to be so off that I never, ever once get the "I could believe it" sensation where some can produce this sensation.

(It also helps to have recorded and played back instruments one is very familiar with on various speakers. That takes a little bit of the variables out where you actually know what that instrument sounds like. Some speakers to my ears produce my acoustic guitar like *something clear being strummed in front of me* where some few produce a timbre that strikes me more like *that's how it SOUNDS in real life* - like the color being adjusted from "off" to "accurate" on a TV or a photograph).

I just came back from a pal's house where I brought over some records I was listening to on my MBL speakers, e.g. a Bruce Cockburn album with a track of a single acoustic guitar piece. In either the "open eyes" or "closed eyes" not for a second did I get the "that could be a real guitar" sensation from what I heard. Whereas the life-like impression and timbral convincingness seemed to come almost effortlessly listening through the MBLs.

It would be great to have a lab where I had both sets of speakers, a recording of an acoustic guitar playing and compare both speakers, blinded, to the sound of the real guitar playing to see if my subjective impressions hold up. But....lacking such facilities, and not having access to any other science (that I'm aware of) concerning blind testing of live vs recorded....for someone who IS interested in experiencing realism, my "can I believe this, is it giving me characteristics of live instruments that sound 'right' to my ears" is the best I've got.

 

[andreasmaaan]

Because microphones don’t pick up interaural timing and intensity differences that our ears do (and even where binaural mics do pick these up, loudspeakers can’t reproduce them), the “truth” contained in an un-EQ’d/uncompressed recording is not the same “truth” a listener at the original acoustic event hears - no matter what steps you might try to take to make it so.

In this context, where a recording is EQ’d or compressed to sound more like the engineer or artist’s idea of how the original acoustic event sounded to them, I don’t think it can be said that the result is further from the “truth” than the same recording prior to EQ/compression.

Of course, a lot of mixing goes beyond the goal of recreating (someone’s idea of) the sound of the original event in the original space, and into the realm of creating a particular aesthetic. But the line is certainly not clear, even if it’s possible to identify examples at both extremes.

I guess my point here is that EQ and compression are not inherently a corruption of the “realism” of the recording, which is in any case elusive from the beginning.

 

[MRC01]

...
When I close my eyes and examine the sound coming from a speaker, it's not that I know exactly what any particular instrument sounded exactly like in front of the microphones. But I DO know, I believe, that they are playing instruments I'm generally familiar with hearing live and that the live instruments would have a distinctiveness that I'm not hearing through the speaker. So it's not so much "I know exactly what those particular instruments in the musician's hands sounded like in real life" but more "I have a good idea by now how instruments tend to sound in real life."
...

Almost no speaker or headphone is so bad that you can't tell what instrument is playing. I know that's an oboe, that's a cello, that's a violin, even on shortwave or AM radio. And we can identify people by their voices on a telephone, which is low-fi. But that doesn't mean it actually sounds like a real oboe, cello, etc. "I could believe it" is a difference of degree, not of kind.
I like Floyd's analogy: we take hints and mentally elaborate them to imagine the event they represent. High fidelity is a more detailed hint capturing more of the musical experience, leaving less to imagination.
With the best recordings and playback systems, the illusion can be convincing. Occasionally I hear stuff in a recording that makes me look around the room to see if that was a real sound.

... EQ and compression are not inherently a corruption of the purity or realism of the recording, which are in any case elusive from the beginning.

I agree. Audio processing or manipulation is a tool, and like any tool, it can be used for good or for bad. But this negative connotation was earned. For every recording engineer who uses these tools judiciously to better capture the musical event, it seems there are 10 others who use it to make stuff sound as loud as possible, squashing the music into gross caricature.

 

[MRC01]

... *Why* is the electrostatic good? The superior phase response or the directionality at high frequencies, etc.? Or could it be psychology; that a visibly-exotic structure like an ELS is bound to inspire a more positive response than a common or garden rectangular box?
...

All possible. Another factor may be that planar drivers are physically large so a given SPL requires smaller movement displacement, which is one of the factors that contributes to distortion.

 

[andreasmaaan]

Audio processing or manipulation is a tool, and like any tool, it can be used for good or for bad. But this negative connotation was earned. For every recording engineer who uses these tools judiciously to better capture the musical event, it seems there are 10 others who use it to make stuff sound as loud as possible, squashing the music into gross caricature

That’s very true. I think there’s broadly speaking a third category though, in which processing is used to achieve a non-realistic goal, but that goal is not to simply make the music as loud as possible. This is what I was hinting at with my use of the phrase “to create a particular aesthetic”.

 

[MattHooper]

Almost no speaker or headphone is so bad that you can't tell what instrument is playing. I know that's an oboe, that's a cello, that's a violin, even on shortwave or AM radio. And we can identify people by their voices on a telephone, which is low-fi. But that doesn't mean it actually sounds like a real oboe, cello, etc. "I could believe it" is a difference of degree, not of kind.
I like Floyd's analogy: we take hints and mentally elaborate them to imagine the event they represent. High fidelity is a more detailed hint capturing more of the musical experience, leaving less to imagination.
With the best recordings and playback systems, the illusion can be convincing. Occasionally I hear stuff in a recording that makes me look around the room to see if that was a real sound.

Yup. And as Floyd says and we've mentioned earlier in the thread, it also depends on the degree we let our mind go and accept the illusion, fill in the gaps.

But moments ago I did another listening session that seems to re-enforce what I've been saying. I'd played the acoustic guitar track on my friend's mega-buck system and though it was clear as a bell and I could hear super clean differentiation between all the guitar strings....it just didn't "sound" like an acoustic guitar I know. The color was "off" it didn't sound real. I went home, played my acoustic guitar and the difference seemed obvious - the types of "tonal colors" I hear from a real guitar just aren't evoked by my friend's system. I played the guitar track through my big Thiels and it was much better, more in the ball park of "the real thing" but then I hauled out a very old pair of Thiel 02 monitors - cheaply constructed, thin-walled little box speakers circa 1980. I've kept them around because they seem to have a "magic" tone especially with acoustic guitar that seems incredibly "right." I played the guitar track through them and...bam!...wow!...that was IT! THAT sounded like an acoustic guitar being played in front of me. That particular tone of the low strings against the wood body, the particular tone of the high strings. It wasn't as super clear, wasn't as smooth and refined as through my friend's system, but everything sounded "made of the right things" the woodiness of the body resonating, the particular tone of the strings. I could sit and listen all day. I picked up my acoustic guitar and played and, yes, THAT's what an acoustic guitar sounds like. It was almost bang on like the real thing.

I just don't know how to account for what it is about certain speakers, like those cheap Thiels, that seem to produce sounds that seem so consonant with what I perceive in real instruments. But I do find that this seems to be very predictive of my enjoyment of a speaker. If it's not there I don't care to sit in front of speakers and listen, may as well be background music. If it's there....I'm transfixed and can listen all night.

 

[Blumlein 88]

Yup. And as Floyd says and we've mentioned earlier in the thread, it also depends on the degree we let our mind go and accept the illusion, fill in the gaps.

But moments ago I did another listening session that seems to re-enforce what I've been saying. I'd played the acoustic guitar track on my friend's mega-buck system and though it was clear as a bell and I could hear super clean differentiation between all the guitar strings....it just didn't "sound" like an acoustic guitar I know. The color was "off" it didn't sound real. I went home, played my acoustic guitar and the difference seemed obvious - the types of "tonal colors" I hear from a real guitar just aren't evoked by my friend's system. I played the guitar track through my big Thiels and it was much better, more in the ball park of "the real thing" but then I hauled out a very old pair of Thiel 02 monitors - cheaply constructed, thin-walled little box speakers circa 1980. I've kept them around because they seem to have a "magic" tone especially with acoustic guitar that seems incredibly "right." I played the guitar track through them and...bam!...wow!...that was IT! THAT sounded like an acoustic guitar being played in front of me. That particular tone of the low strings against the wood body, the particular tone of the high strings. It wasn't as super clear, wasn't as smooth and refined as through my friend's system, but everything sounded "made of the right things" the woodiness of the body resonating, the particular tone of the strings. I could sit and listen all day. I picked up my acoustic guitar and played and, yes, THAT's what an acoustic guitar sounds like. It was almost bang on like the real thing.

I just don't know how to account for what it is about certain speakers, like those cheap Thiels, that seem to produce sounds that seem so consonant with what I perceive in real instruments. But I do find that this seems to be very predictive of my enjoyment of a speaker. If it's not there I don't care to sit in front of speakers and listen, may as well be background music. If it's there....I'm transfixed and can listen all night.

Hmmmm, I've noticed this before with guitar recordings. I think that small monitor being of a similar size and volume as the guitar helps somehow.

 

[Cosmik]

Finally, to expect anything resembling perfection from a two channel record/reproduction system is a fantasy

I don't see that as a given. We only have two ears, and so two channels is all you really need. What comes over the channels and how they reach your ears is the question.

Twenty seven channels is not much better than two channels compared to the infinite number of channels that perfection implies. And then twenty seven channels seems a bit 'blunderbuss' in comparison to the two-channel system that I, and some others, are often fascinated by.

My suspicion is that surround sound is a ho-hum 'ambience button' while stereo is a remarkably exquisite 'hologram'. Of course you can have the hologram and the ambience button too, I suppose.

 

[RayDunzl]

stereo is a remarkably exquisite 'hologram'.

Stereo is not a hologram.

I've played with a real hologram.

It's spooky.