Are Loudspeakers Musical Instruments?
parts of an e-mail correspondence, conducted on the Sursound listserv and elsewhere

Subject: RE:sursound: Realism in audio reproduction...

When we engineers have finished our job and the final product is ready to be heard, what exactly IS it that emerges from the loudspeakers? Is it a realistic portrayal of some music making that happened somewhere else?

Well, the latter appears to be one goal, certainly. Is it musical? It had better be, or there was no point in doing all that work! Is it "real", in the sense of sounding exactly as it did in the hall? Of course not!

For a host of reasons, the reality of the classical music acoustical recording session becomes, in a recording, a deliberate and choice-laden translation from one medium (live music) to another (loudspeakers) and from one acoustical environment (the hall) to entirely different places containing loudspeakers which along with the rooms they're in add their own acoustical signatures to what is heard - that's simply unavoidable.

Further, our choice-laden path does not tend toward the direction of the greatest accuracy or precision of representation, but toward the representation of what "sounds best." I can't imagine saying to a client (and thereby committing professional suicide), "Look, I know it doesn't sound as it good as I can make it sound, but this is as close to the way it happened as we can record, so I refuse to improve it by adding equalization, reverb, editing, or level changes." Such an approach would be, to put it gently, the wrong paradigm!

The loudspeakers have replaced the original instruments, becoming now the musical instruments themselves, playing note-for-note a performance that was once played somewhere else. That being the case, perhaps the most important question that needs to be asked about that recording is whether or not it was crafted well enough to survive the journey with as little harm as possible being done by a new set of instruments in a new room to the musical instructions it contains. The better the instructions, the better those different loudspeakers - and the room they're in - will be able to follow them.

David Moulton, who is both a musician and a recording engineer whose interests encompass both arenas speaks eloquently and often about loudspeakers as musical instruments. A sample of those thoughts can be found in a recent post to the pro-audio internet forum, in pro-audio-digest V1#163, 5 Oct 1997 under the title "Re: Speaker philosophy'n musical instruments'n stuff.

Copyright 1997 by Curt Wittig, Washington, DC
----------------------------------------------------
For those of you who might be interested, here is what Dave said on the ProAudio listserv:

David Moulton:
(posting to ProAudio)

After a week of thinking about it, I guess I've got some things to say about the series of posts regarding loudspeaker philosophy, theory and practice, including the fairly obstreperous ones from Robert Greene and others from Manny LaCarrubba, Jim Johnston and Bob Ohlsson. Sorry, but this is going to go on for a bit.

Manny LaCarrubba cited me in regard to the notion of "speakers as musical instruments," a concept that Robert Greene has dismissed in less than kindly terms.

A little history and context here. I first ran across the idea of "speakers as musical instruments" long before I got involved in audio, in fact as a graduate student at Juilliard in the early 60s. Vincent Persichetti pointed out in a seminar on 20th Century music theory that any speculations we make about the sounds of musical instruments based on our listening to recordings were not going to be reliable. He pointed out that when we listened to recordings we were listening to "loudspeaker music," and that such music is significantly different from orchestral music or piano music or string quartets or whatever. Persichetti stated (and I now believe that he was correct) that loudspeakers have a distinct sound quality as a family of sound-producing devices, just as pianos, guitars, violins and brass do.

It is also worth noting that it would be difficult or impossible to construct a physical test for distinguishing musical instruments from other devices that would include the tympani, the pipe organ and tuba in the set of musical instruments but exclude the loudspeaker.

It helps, in both the production of recordings and the design of loudspeakers, to keep this "musical instrument" perspective in mind. Once we accept that absolute realism is an inappropriate goal (more about this in a moment), we can devote our creative energies toward the evocation of what I call the "ness" of a recording. That "ness" (orchestral-"ness" or piano-"ness" for instance) is an evocation of the musical spirit implicit in the performance of music via an orchestra or piano (for instance), TRANSCRIBED for loudspeaker. At the same time, it is an evocation of the musical spirit of the performing artist(s), an evocation of their musical personae AS STRONGLY AS POSSIBLE! Such an evocation is at the heart of great recordings. (If this sounds hopelessly fuzzy, romantic and new-age, please take a look at my paper in "The Sound of Audio" (AES) entitled "The Creation of Musical Sounds for Playback Through Loudspeakers," for a more detailed and clinical exposition of these ideas.)

This is different from the goal of accurate sound reproduction. Accurate sound reproduction, taken literally, means as precise a replication as possible of the acoustical events that occurred in a given space at a given time. The test for quality of replication would be a comparison of the original with the replication. In such a case, of course, ANY variance from the original physical acoustical situation would be bad, and so devices such as attenuators, equalizers, editors, compressors, reverbs and (dare I say it?) transducers should probably be regarded as bad devices. Accuracy implies the least possible change, and for our work, that would include the least possible change (both relative and absolute) in sound pressure, sound spectra, sound direction and sound time.

Yet even the briefest look at our current recording production practice makes very clear that we casually change all of these things as a matter of course. And we do so for a very good reason. Our goal is to make the MOST BEAUTIFUL recording possible, not the most accurate one!

In the verbal presentation of the above-mentioned paper on the psychoacoustics of multitrack recording for the AES Sound of Audio Technical Conference in 1990, I suggested that many bands have a bass player named Joey, and that an accurate rendition of Joey's performance is probably the LAST thing we want to do. I suggested that it would be professionally irresponsible to insist that we NOT try to enhance Joey's marginal performance and sound just because to do so would be a more accurate representation of what really happened in the session. It would be equally professionally unthinkable, for instance, to refuse to equalize out air conditioner rumble on the grounds that to do so would be a less accurate representation of the acoustical event that was recorded.

What does this have to do with loudspeakers? Plenty. If there is something to the concept of high fidelity (and I believe that there is), it comes from (a) a coherent relationship between what the producer approved when he or she listened to a playback of a master recording and what the listening public hears when they play back a released copy of that recording, and (b) the richest possible sharing of sonic information gathered in that recording with the listeners.

(a) requires that producers (and their engineers) anticipate what recordings WILL sound like for end users - I call this Listening Ahead Monitoring (Listening Back Monitoring occurs during tracking sessions, when we need to assess the relationship between the acoustic performance and the recording). This involves considerable review and test listening in a wide varieties of environments. As my neighbor Tom Bates put it, "the engineer isn't done mixing until the recording sounds good on all of the loudspeakers available, not just his or her favorite." We test in cars, on Auratones, over headphones, in mono, and at a variety of levels. Why? Because the end-user range is huge. There is no single end-user specification, only a range which includes the following FAMILIES of playback systems (God help us!):

• stereo living-room systems,
• mono table-radios/TVs,
• headphones,
• cars,
• boomboxes/stereo TVS/multimedia-spkrs,
• boom-vans, and
• home theatres.
  

Faced with this multiplicity of families of systems, plus the large range of variability within each member system, it simply is not possible to predict how any given playback will sound, as Robert Greene so fervently desires. There is no "reciprocity paradigm" (Greene's words) in place. Even worse, this is before we begin to consider the effect of the playback room(s) on the playback over any of these systems.

So, as a production practice, producers try to anticipate and make recordings that will "work" over their particular target range of listeners' playback systems. And consumers (and their suppliers) who are concerned about this try to come up with systems that are reasonably similar to what they believe the producers used when they signed off on the master. This is, happily, a self-reinforcing, positive feedback loop (which includes the mastering engineers who try to correct, as "ears of last resort," any defects they perceive in the master), and as a production practice it works reasonably well.

The second thing we try to do, (b), is give the listener as much valid recorded information as we reasonably can. Happily, stereo gives us at least a little directional information, which it turns out that we, as listeners, both like a lot and find very useful for gaining some additional information both about the performing instruments in a recording and the space(s) in which the recording was made.

Now then, a primary reason why stereo works is because of the precedence effect ("the law of first wavefront", if you will). On the median plane, the precedence effect is defeated and so, of necessity, we mentally fabricate an imaginary phantom image whose location best fits the available data, which includes at least two early reflections (the sounds from the two loudspeakers). This is true not only for the direct sounds in a recording, but also the early reflections and reverberance in the recording. Our localization mechanism allows us to fabricate both a sense of source and a sense of space from even the very lossy data presented to the two phase-locked loudspeakers.

Jim Johnston points out (citing Fletcher and others) that if the angle between the two loudspeakers gets very large, the phantom image illusion falls apart. Based on my research, I believe this happens because such wide angles for first early reflections, unsupported by either an actual direct sound or other appropriate early reflections from the playback room, imply a location that our cognition "knows" is not true.

So, on the one hand, Greene would like us all to move our loudspeakers to a 90° spread to listen to Blumlein recordings (based, I assume, on the +/-45° placement of the coincident microphone pair), while Johnston points out that if we do so, we won't get a reliable phantom. Meanwhile, Green would like us to suppress the reflections of the playback room as an undesirable "acoustical signature" that masks the "acoustical signature" of the recorded room and performance. He would have us listen out of doors or in anechoic chambers.

Well, I've gone to the trouble to sit down and listen at length (20 hours or so) to stereo recordings over a variety of loudspeakers in an anechoic chamber. While the experience was an extraordinarily interesting and informative one, it did finally seem to me that the anechoic chamber was neither a useful place in which to prepare recordings for release nor an "accurate" environment in which "it [would be] possible to come very close to defeating the sense that one is in the acoustic environment of one's . . . listening room and creating in its place the feeling of being in a different acoustic environment altogether that resembles the original site of the recording" (to use Greene's words). While that goal is quite laudable, the simple direct information from two loudspeakers running in stereo simply does not provide enough information to satisfy our hearing's voracious appetite for complex input - we know, from the LACK of data as much as anything else, that the stimulus is bogus. That lack was sure audible in the chamber.

This is where wide dispersion loudspeakers and early reflections come in. In fact, as noted by several posters, we cannot eliminate low frequency reflections in rooms except by extreme (anechoic treatment) measures. Further, we cannot prevent long wavelengths (2x the largest dimension of any given speaker driver) from radiating omnidirectionally into the space. Given these physical constraints, our attempts to control the directivity and early reflections of loudspeaker sound in rooms has resulted essentially in the practice of severe and erratic low-pass filtering of the reflection paths in the playback room. It is this erratic off-axis behavior that makes loudspeakers with similar on-axis response sound so different. Meanwhile, our auditory system uses the high-frequency spectra of sounds to localize their sources in a reverberant space, utilizing all of the phase-locked volley of direct and early reflections of a sound, to establish (a) the location of a sound and (b) its timbre. WHEN we employ a loudspeaker that has both point sources and constant lateral output as a function of frequency, working in a room with hard sidewalls for broadband specular reflections, it turns out that the resulting playback phantom images and reverberance GAIN in detail, not LOSE. Thanks to our hearing system's ability to integrate the early reflections and the direct sound into a holistic perceptual construct (derive coherence from chaos, if you will - this is what we do ALL THE TIME with normal acoustic sources in rooms, with no trouble or even any conscious awareness), the early reflections of the playback room carry useful auditory information ABOUT THE RECORDING to the listener. Conventional loudspeakers fail to deliver this information, for which our ears are starved (cf. Benade).

This isn't just theory (although I have several times over the past decade stated the underlying reasoning in writing for such a theory). It is also successful working practice. At present, I am using loudspeakers (designed by Manny LaCarrubba, incidentally) that are essentially flat from 30 Hz. to 17 kHz. over AT LEAST 180° laterally and 30° vertically. I use them in a large (8000 cubic feet) reverberant (.5 sec. RT60) room for production purposes (production, postproduction and mastering, plus research) as speakers of choice. They are configured as a 6-channel discrete playback system (including center and overhead speaker). Because of this configuration, the angle between L and R is in fact 90° at the sweet spot, and there is also a center speaker in place.

Let me share an anecdote: recently I was visited by the senior loudspeaker designer for a major loudspeaker manufacturer who spent two days listening to these speakers in my studio. At one point, we were listening to a mono piano recording while the engineer was sitting in the sweet spot. He found the mono phantom generated by L+R to be so realistic that he finally, with apologies, had to verify for himself that the center speaker was turned off by getting up, walking up to the center speaker and putting his ear up to the tweeter lens. One of his final comments was, "I can hear no problems whatsoever with the speakers and no problems with the room." (Remember, 90° spread between speakers, big reverberant room, extremely wide dispersion - should be a nightmare, right? Well, it definitely isn't! The imaging is regularly cited by clients and visitors as being remarkable.)

So, Robert Greene has a point. 60° may simply be a convenience for more directional speakers, and a common practice evolution that has no particular psychological or physical significance.

BUT, Greene's argument against the use of broadband early reflections in the playback room fails to take into account the nature of the precedence effect and its function in our auditory localization system. As a result, he would substitute bad data (low-pass filtered early reflections) for good data AVAILABLE IN THE RECORDING. This is a common-sense, reduce-the-variables, intuitive approach (I used to believe the same thing until I got to mucking about with omnidirectional prototype speakers), based on an incomplete understanding of our hearing mechanism and its ability to resolve chaotic signals.

FWIW, I've been employing design principles based on this reasoning for studio design for the past several years with considerable success - which is to say that the clients enjoy the rooms, enjoy working in them and are satisfied with the recordings they produce working in such rooms. I wrote an article on such rooms for the April 97 issue of Recording Magazine, and I've presented a couple of papers at AES on related work (LA, 1986 and N, 1995), if any of you want to go farther with this.


Enough. Thanks for reading all of this.

Dave Moulton
Copyright 1997 by David Moulton Prefessional Services, Groton, MA


David Moulton Professional Services
and Digital Media Services
Groton, MA 01450
978-448-6828, fax: 448-6851
dmoulton@ma.ultranet.com
Copyright 1997 by David Moulton Professional Services. All rights reserved

This is further to the perspectives set forth by Curt Wittig and the posting he cites by David Moulton to the ProAudio listserv, thinking of loudspeakers as musical instruments with their own natures; as translators of, but by no possible means absolute replicators of, other ("real") instruments and acoustic spaces. I agree with them.

But I'd like to add or at least to reinforce the point (which I doubt they dispute), that the effort towards perfect fidelity is well worth making. Many listeners seem to find in their experience of surround recordings, as in high fidelity recordings, a fascination with the space in which a recorded performance takes place. They value a delivery to our ears -- loudspeakerness notwithstanding -- of clues that contribute to an illusion, however imperfect, of "being there then and now" as the music was being performed by the musician(s). It's an illusion not just of somewhat inhabiting the performance space, but also of being in the presence of the actual instruments in real time. Listeners to surround recordings sometimes comment on the instruments' pleasing three-dimensionality -- a related impression of realism. Those illusions, products of striving towards accuracy, are surely one of the colors (for want of a better word) that producers and engineers may use to convey what David Moulton calls "the musical spirit of the performing artist(s)."

Not to reduce the loudspeakerness vs. realism argument to a pirouette upon neutral ground, but perhaps nobody strongly disagrees that we must have it both ways. Striving after accurate sound reproduction helps the cause; gives the producer and engineer valid qualities to evoke from their and their listeners' loudspeaker "instruments". It is one of a set of tools to use -- effects or other distorting devices of all sorts can be other tools, please note -- to effectively translate or re-create the musical spirit and beauty of the artist's work.

That leads me to another word about loudspeakerness, and about the objectives of producers' and engineers' work. As long as loudspeakers are part of what all listeners listen to, it is important that some of us acknowledge, study, and work with their nature. There are parallels in other arts. Words, for example, are signs of wordless as well as verbal consciousness, but perhaps none of us would venture that written or spoken words are the same thing as wordless consciousness. (We might be wrong about that, but that's another discussion.) And painting, sculpture, dance, film, theater and other arts incorporate or plausibly represent "real", familiar gestures, shapes, images, actions, etc., which give us great or subtle shocks of recognition, all the while adding their own terms or signatures.

So it is with loudspeakers. They deliver their own artifacts/character as instruments even as they may deliver worthwhile illusions of reality. One way or the other we often call the best of them transparent, sometimes even better than or more real than real. "More real than real" is what the person said who, after being visited by an angel, could not doubt the visit but still found it hard to tie into his "normal" realities. Angels, as Chesterton once said, are beings who take things lightly. Maybe we mortals tend towards the ponderous, as if our living and dying were momentous. At any rate, each of us in our own way (some shareable) knows truth, beauty, and meaning for what they are. We know music to our ears when we hear it. As conveyors, translators, creators or re-creators of sonic art, we work with such dynamics, with forces both ordinary and sublime. It's good work when we can get it, eh?

Very lively, it has become, yea, verily, the List has been enlivened beyond the limited radius of its earlier range of verbiastic persiflage....

Since I have joined the Orders and sold everything except my wire recorder, I cannot add to the intellectual potage at this time, but I did want to convey my appreciation....

I have some ideas regarding loudspeakers as musical instruments, but they are still at the preconscious level and my Abbott would have to release me from my vows of silence in any event.

Curiously enough as an instrument builder I have participated in eerily similar discussions. Six months to a year spent on creating a harpsichord along some famous elder guidelines would create what some would call only a transparent medium through which the essence and spirit of the music should be transported to the listener's acoustic realm.

Others would get completely wrapped up in the "character" of a particular instrument, seeking to bend the musical interpretation to the nature of the instrument; and curiously enough either approach, if well executed, would lead to an exciting performance, with unexpected musical qualities and insights.

With so many different national and regional schools of instrument building - French, English, Flemish, German, Italian, and so on - all serving different musical expressions, from Frescobaldi to Bach and Rameau, with their varying demands and sounds, one might say that as an instrument maker I was usually called upon to create something with very specific lineage and character, as far from being a transparent medium as one could imagine, and yet:

Some instruments might not be as tractable as one would hope, or perhaps of the wrong school to serve a particular musical idiom, yet some performers could elicit miracles from them. Other instruments might fly like the very wind and carry forth the spirit of the music with perhaps only the slightest effort by the player; the right combination of touch and projected sound might also serve to "teach" - or shall we say serve - the player during the actual performance, and bring out musical nuances which enhance the poetry of the event.

Various players have occasionally commented on this "teaching" aspect, stating things like "I didn't know the music could be played that way" or "I'm hearing things I didn't even know were IN the music" - not the sort of reaction one would expect from a "transparent medium".

Let's return to the notion of loudspeakers as musical instruments: If you consider the above, which represents my experience over 20 plus years making instruments and dealing with concert artists, then I can't really think of speakers as musical instruments. Obviously, they all have their specific and special qualities and are built with the care, thought, and artisanry which I know well as a keyboard instrument maker; so I don't discount that in the least. But I do see the aim as different: to create a transparent medium which serves the original performance as best it can, rather than to color and shape in a manner not present in the original. As a harpsichord maker, I sought deliberately to color and shape. Violin makers do as well, I know.

Mark Parsons:

To develop the SPEAKERS AREN'T MUSICAL INSTRUMENTS case a bit more, especially in light of your wonderful remarks about qualities of harpsichords, Yves, there's plenty about calling loudspeakers musical instruments that rings untrue. For one thing, next thing we know everyone else in the creator-to-listener chain may claim to have an instrumental hand in musical performances. We'll be asked to applaud the artistic contributions of the makers of speaker cables and hi-fi electronics and even the binding posts attached thereto, the makers and marketers of CD's, manufacturers of the air conditioners that alter the air pumped into the world's listening rooms, contractors and subcontractors of same rooms, drivers and manufacturers and designers of passing trucks whose continuos may intermingle with those emitting from our woofs and subwoofs, even pro audio sales weaselships like mine which may somewhat influence users to choose one instrument/tool instead of another. {;-)

In some important ways soon-to-be-obsolete loudspeakers are not the same as Strads and Strats and wondrous harpsichords. Those differences are well known to us all, I'm sure, and perhaps have fueled your prior reluctance to call loudspeakers (marvels that they be) what they aren't quite. In good measure you were right that, unlike true musical instruments, loudspeakers are one of the things that should just stay out of the way of the music or, as you put it, should provide "a transparent medium which serves the original performance as best it can, rather than to color and shape in a manner not present in the original."

On the OTHER other hand perhaps everything in the chain, including the musician and his gorgeous (i.e., colorful and coloring) axe, is supposed to expand upon but not impede that Originality. Frank Sinatra has been quoted lovingly calling his microphone an instrument. Most musical artists who perform in public necessarily treat everything within earshot as their instrument, all at the service of the Original. What musician who wishes to record can fail to wish all possible artistry applied to the full path that his music travels, including many an electronic device? Accordingly the playing of the harpsichord, most instrumental, may constitute most of that journey, but not all of it.

Curt Wittig:

Musical instruments are speakers!

But that's just my opinion. And it could change at any moment, having only been arrived at a few seconds ago. ...

To be honest, I confess to not really caring very much whether or not anyone thinks that loudspeakers are musical instruments, but only that peoples thoughts on this subject suit their needs. In my case, until recently I've never thought about this subject at all, though now that it's come up I suspect I've probably been using loudspeakers as musical instruments all my life and so it doesn't really feel strange talking about them that way. Have I spent too much time thinking about something I didn't really need to think about at all because whatever conclusions I came to wouldn't have made one bit of difference to how I use loudspeakers anyway? Sobering thought.

But maybe in the end it will turn out that loudspeakers are merely loudspeakers. Geez, I hope not!

END

to top of page

back to learning

Copyright 2002, Parsons Audio. We welcome your questions, comments, and contributions >>> Webmeister@paudio.com.