John Whiting takes us on a whirlwind tour of mono,
stereo, surround, and Ambisonic sound
It's like a story from Weird Tales: "Ambisonics – the Technology that Refused to Die!" I've listened to and worked with so many enthusiasts over the years, starting with Peter Fellgett, Geoffrey Barton, Bob Woolford, and Michael Gerzon; and great musicians such as Andrew Parrott, who was exposed to the virus at Cambridge and until fairly recent years nursed an ambition to record the Josquin Notre Dame Mass, singing all the parts himself one by one, digitally overdubbed, from plotted locations around a stationary Soundfield mic. Along the way there were almost-triumphs, such as the Great Boots Breakthrough, in which Britain's leading "drug store" was to bring out a cheap, compact Ambisonic system to sell along with their mini-cameras and hair dryers.
Ambisonics is organically related to, though not derived from, an already-existing technology – M-S recording as developed in the early thirties by A.D. Blumlein, the English scientist who was also responsible for the 45/45 stereo record cutting technique. M-S configuration has been repeatedly demonstrated to achieve the highest accuracy of stereo imaging with two loudspeakers. This was made obvious in experimental recordings made by James Boyk at Caltech in 1988, in which eighteen different stereo mic arrays were used to record an electronically generated tic played successively across a line of two-inch loudspeakers nine inches apart. One may argue as to whether low-frequency information should have been tested as well, but within the experiment's chosen parameters, the reaction of my classes in recording techniques has always been unanimous.
Some of us had always dreamed of the wide open spaces. In the old mono days, the more phase-coherent a single loudspeaker was, the more the sound seemed to come through a tiny hole in the wall. Using more than one system, if they were truly compatible, merely moved the hole to a phantom spot midway between them. I well remember the excitement of first hearing the antecedent of the Bose systems, the "wall of speakers" in which you cut holes in a large sheet of plywood to accommodate as many small, cheap units as you could afford, wired up in both series and parallel so as to present a sensible impedance to the amp.
The theory – and it worked pretty well – was that defects in each speaker would average out and be covered by the others: a system designed not by, but like, a committee! It was, in fact, wonderful to listen to an organ which ambiguously filled the space in pretty much the way it filled a church, particularly if your listening area (in my case, Spartan and academic) was generous and not deadened by wall-to-wall carpeting, heavy drapes and overstuffed furniture. The important thing was that, although the sound came from in front of you, it appeared to cover an area commensurate with the size of the original source. Further variations involved the use of speakers around the space, in or out of phase, but the effect was most satisfying if there was some kind of frontal predominance such that you weren't simply immersed in a Jacuzzi.
Then came stereo and, once we had outgrown the ping-pong and the passing locomotive, we discovered that space could be solidly defined in front of us, due to the fact that instruments at different distances from a pair of mics (whether spaced or coincident) were captured with different ratios of direct and reflected sound. The peep-show had become not a mural but a sculpture, or at the very least a bas-relief. We had discovered, serendipitously, the beginnings of a working methodology of aural perspective. The challenge was similar to that facing the renaissance artist: using two dimensions to create the illusion of three. (Forget for the moment that stereo is really only one dimension and flat surround only two; because of reflected sounds, most of us don't "visualize" it that way.)
In the early days, all stereo recordings, whether spaced or coincident, were made in remarkably similar ways. When we heard instruments coming from different places, some of us began to be much more critical of location than we ever were in a real concert hall, where the evidence of our eyes automatically took precedence over that of our ears. Some of those using spaced omnis became aware of the "hole in the middle" and compensated with an extra central mic fed equally to both channels. This helped somewhat, and did not violate the relative depth integrity which was inherent in a single line of mics.
Pop music, in the meantime, was evolving a totally different way of working, which came eventually to require an infinity of souces and an infinity of tracks on which to record them. This was partly due to the fact that all the musicians could not be relied upon to play all the notes correctly at the same time.
Then came the inevitable moment when an accountant wandered into the studio during an orchestral session. The next day he summoned the producer into his office. "What's the big idea?!" he yelled. "All those musicians, all those mics we just paid for, that new Ampex multitrack, and you're only using two old ribbons and a clapped-out 351-2!"
Thus multi-mic orchestral recording was born. Stereo recordings became murals on the wall, every instrument equidistant from its audience, because all the mics were arranged, as much as possible, to hover over a single sound source. We listened with a hundred ears, each focused on its own segment of the sound picture. It was like looking at a primitive painting in which each area of the canvas is seen as if the eye were immediately in front of it. We had gone back to a pre-Giotto mediaeval perspective.
There was something to be said for this approach, particularly when listening to music on tiny systems with a lack of definition. Decca "Phase Four" recordings, with the appropriate mic lifted for each solo passage (what I have elsewhere characterized as "the phenomenon of the ten-foot piccolo") sounded grotesque on large stereo systems, but were remarkably effective on Granny's tranny; on fuzzy old consoles, instruments no longer sounded as if they were calling for help from the bottom of a well.
Recording techniques in the abstract, however, are of much less significance than the specific microphones and loudspeakers which are used, together with their placement in the particular spaces in which they are employed. The location of speakers as determined in most home environments makes a nonsense of all the care which may have gone into the recording. It's like smothering tournedos Rossini in HP sauce.
In a typical living room, where cosmetic criteria are paramount, stereo speakers must be small and tucked away unobtrusively into the decor. This means that one speaker is liable to end up on top of the liquor cabinet and the other behind the sofa. It was the acceptance of this inexorable scheme of things which gave Bose in America a distinct edge in the domestic hi-fi scene, with its development of the rear-firing jet-propelled 800 series which spread the sound about the room like LA smog. With such an amorphous sound source it's no wonder that spaced omni recording, with its full-frequency bass response, became the favored stereo method, even though when such recordings are heard on a properly placed and balanced system, soloists are apt to leap about the sound stage with an athletic agility which would earn them Olympic gold medals!
Thus, when the alphabet soup of quadraphonic systems came along, they hit a market in which there was by no means a consensus as to what a stereo system on its own should accomplish, let alone the extra speakers at the back. Those who just wanted to spread it around a bit were happy with whatever they got, while the super-critical listener who wanted to be able to draw lines on the wall to mark first and second reflections soon found that none of the systems were up to it. (There's an old Peter Arno cartoon of a little mustachioed pedant sitting in a night club pointing at the last naked girl in the chorus line and proclaiming, "She's an eighth of a beat out.")
A lot of air has been heated over the compatibility of UHJ-encoded Ambisonics with conventional stereo. How many degrees off-axis can you go before the image starts to shift? In my own experience there has been a general consensus, after A-B tests, that UHJ recordings are about as stable as good M-S stereo recordings made under similar circumstances, which is pretty good. In fact, if you play a UHJ recording through a decoder and listen to the front channels, you are hearing the rough equivalent of an M-S recording with a sub-cardioid mid-mic.
Note that I have specified the front pair only. UHJ-encoded recordings heard undecoded in stereo must inevitably superimpose the rear signals on top of the front. Thus a generously ambient recording which sounds wonderful when you're in the middle of it can, in stereo, unceremoniously dump you at the back of the hall. This is why Nimbus have had to live for years with a reputation for producing wishy-washy recordings. Some of their later efforts, such as the Rozdesvensky Stravinsky recordings at Watford Town Hall, were done from a closer perspective which compensated for this. But everything has its price; hearing them on a good Ambisonic system, the listener is ensconced on the conductor's podium. Great for closet baton wielders.
Within this context, discussion of the precise placing of four or more loudspeakers in a domestic environment becomes as abstract as the mediaeval argument over how many angels could dance on the point of a pin. If surround sound takes off as an adjunct of new video technologies – the only circumstances within which it might become commercially viable – the rough-and-ready distribution of Dolby Surround will be more than adequate for its great new public, as it already is for those who sit in cinemas, where its anomalies are much more evident. Did you ever hear anyone mutter on the way out that the sound stage wasn't very accurate?
The situation is complicated (or simplified, depending on how you look at it) by the fact that the directional precision of our hearing is greatly reduced to the rear; therefore, accuracy of assignment to the rear speakers is not nearly so important as to the front, particularly if the source material consists of sounds which are confined to a frontal stage. Even precisely located rear sounds are not accurately detected. Just observe the audience at a performance of Charles Ives' "The Unanswered Question", in which the instruments are placed around the hall, and watch people turning in contrary directions to find the players.
But in spite of the dilemmas on whose horns Ambisonics finds itself impaled, its fans, once addicted, move into a heady stratosphere in which mere stereo becomes a dull flat photo on the wall. Acquiring the paraphernalia with which one must entertain such a demanding mistress can lead the impecunious to the bankruptcy courts. As for the software, there is a scarcity of UHJ recordings which makes the Apple Mac section in a computer store look like Santa's sack. There of course lies the crunch: few collectors can afford to hire the Berlin Phil and stick up a Soundfield.
Nor is the situation likely to change. Behind the triumphant inventiveness and disastrous promotion of Ambisonics lies the same egotistical amateurism which, as if in a Greek tragedy, has followed the rise and fall of one British industry after another – motorcycles, sports cars, radar, hi-fi, computers – each in turn given its inventive impetus by brilliant eccentrics who were unable to bridge the chasms between conception, execution and distribution. When the National Research and Development Council, the government quango which backed Ambisonics' core of inventors, had its opportunity to sell the technology to a Japanese manufacturer, the latter's representatives were treated with such haughty disdain that they went away in confusion. Ambisonics was by now an aging beauty with an irascible guardian; it was amazing that there was still an occasional suitor. Ultimately, in the government's everything-must-go sale of the family silver, the Ambisonic patents were knocked down to Nimbus, which, being the creation of an expatriate middle-European nobleman, was even more lofty and condescending than the civil service.
The small clutch of Ambisonic recordings that continues to appear on other labels is due to the dedication of a few individuals who bypass the normal channels. Ten years ago I was able to slip a UHJ recording of Electric Phoenix, complete with Ambisonic Logo, into the EMI catalog. Just this year Dave Foister of the Guildhall School of Music and Drama has seen his three-year-old recording of Benjamin Britten's Curlew River emerge triumphantly on Koch International, largely through his own persistent efforts. In March it was given a prestigious launch party with an excellent scholarly introduction by David Mellor (yes, really, that other David Mellor!) and a brief sample in an eight-channel Ambisonic playback provided by your "legendary" author. On this occasion I learned how to empty a hall. Simply announce, after the champagne, "Please take your seats. An Ambisonic playback of the entire opera will begin in five minutes."
Only the grimly determined, such as Nimbus, persist; and even they have reduced their Ambisonic evangelism to an unobtrusive message on the disk, like a government health warning. But still a few of us nurture Dreams of Glory in which four whole channels on the nascent video disk will be taken up with Ambisonics, even though no major manufacturer of domestic equipment is backing it, and the encoding and decoding would require hardware which exists only in the museums of a few aficionados. Next week, steam trains in the channel tunnel!
The true probabilities emerged at the recent AES conference in London on "Audio and the New Media", during which some of the world's finest theoretical sound engineers presented a series of papers on surround sound and its possible incorporation into TV transmission and the new video disc. With eight channels of audio available on the disc, Ambisonics champion Dave Malham of York University argued bravely for a "straight through" approach which would at least leave open the option of which system was finally to be adopted.
There was much generous and open minded discussion of alternatives; but at the end of the conference, there was no consensus as to whether the public would buy surround sound, how many speakers should be incorporated, what system of encoding and decoding should be adopted, whether the outputs should be straight-to-speaker or require further processing, whether and how much data compression would be tolerable, and finally, whether the whole thing was worth bothering with anyway. The principal excitement had been generated the previous day by a couple of young Nick Leesons who had cobbled together the cheapest available technology to corner the market in the Internet transmission of countless kilometers of the world's dullest music - never mind the quality, feel the width! Meanwhile, the binding decisions on surround sound will probably be made somewhere on a golf course by men who haven't read the papers or even heard the alternative systems.
Fortunately for most of us, there comes a moment at which we stop listening to the hi-fi and start listening to the music. If we want a system which gives us full-frontal fidelity (to coin a phrase) and a reassuring ambiance, UHJ Ambisonics will provide it as well as any other system and, with some sort of "stereo enhance" processing, even achieve it with non-UHJ recordings. If 360-degree coherence is demanded for sound projection in a large space, with a minimal awareness of individual speakers, B-format 8-channel (or more) Ambisonics probably does it better than any, with the possible exception of the odd one-off prototype.
Sound systems, from mono to periphonic, all rely on expectation and education. Understanding media, whether simple or complex, means understanding "languages" which must be learned. Ironically, some of the finest musicians I know are content with terrible sound systems because they are painfully aware that state-of-the-art electronics does not measure up to a state-of-the-art Stradivarius. Confronted with any recording, they listen to what's there and fill in the gaps. Never forget that contemporaries of the Edison cylinder, who were not ignorant fools, exclaimed at its "amazing lifelike realism". Come the millennium, when stimuli are injected straight into the medulla oblongata, all our systems will be rubbish. In the meantime, our Ambisonic Princess, though imprisoned in obscurity by her cost and complexity, is still kept alive by the magic of her extraordinary beauty. Perhaps some day her prince will come and she can wave a fond farewell to her cottage full of earnest little admirers.
©1996 John Whiting. May be reproduced electronically in its entirety including this message.
Audio Media, Issue 70, September 1996