Undeserved Bad Rap for Ambisonics

To: The Editor,
Home Theater Magazine

Dear Sir,

I was dismayed to see in your March 1997 edition (A New Kind of Channel Surfing, p61), another attack on Ambisonics from Mr. Daniel Sweeney. I wouldn't mind so much if this was a one-off occasion: but in fact Sweeney continually misrepresents and deprecates the system in a way that is obviously deliberate. I really can't understand why he feels so threatened by a surround-sound system!

I won't make a meal out of the minor errors in the article, such as the spelling of Robert Margouleff's name or the mangling of the name of the record company Unicorn-Kanchana. But let's get the facts straight about Ambisonics:

• Ambisonics is not a "matrix" system, any more than, say Dolby Surround or DTS. If you have a 2-channel transmission medium such as CD or FM radio, and you require compatibility with standard stereo and mono, then you are obliged to use a matrix of some sort to encode the surround information into two channels. Dolby and DTS use matrixing systems, and it is entirely legitimate for Ambisonics to use one too in the 2-channel environment. It's called "UHJ".
  
• Ambisonics is not a "remnant of the quad days". It was developed largely independently of the "quad" systems existing in the early Seventies (the encoding matrix mentioned above, which is concerned with the transmission of surround information rather than its implementation, owes something to UD-4, one of the lesser-known "quad" delivery systems). Ambisonics is based on fundamentally different principles. The "remnant of the quad days" that Mr. Sweeney is thinking of is in fact Dolby Surround, which owes its origins to SQ and QS, neither of which worked particularly well.
  
• It is irrelevant to complain that Ambisonic recordings "feature very low separation between adjacent channels (only 3dBs [sic] max)" as the system is not based on the same concept as a "quad" recording. First, 3dB of channel separation is as much as you will get from any matrixed surround system -- for example Dolby surround without logic decoding. More importantly, Ambisonics seeks to capture or synthesize the soundfield -- the overall pattern of sound waves -- present at an event, rather than representing it as four (or whatever) so-called "discrete" individual sources at the points of a square, which is not in fact an accurate method of representing an acoustic phenomenon (that's why there is a center channel in 5.1 systems: it doesn't work very well without one. In fact there are holes between all the speakers). The pattern of sound waves present in a real or synthesized acoustic environment does not respect artificial separation criteria any more than our ears do. Our experience of an acoustic environment depends to a large extent on the relationship between the sound waves arriving from different directions, not on their separation. As in life, so it is in Ambisonics, which utilizes both phase and level information to recreate the soundfield. The relationship between the information channels is vitally important. Throwing it away, as a system which relies solely on level for localization -- such as any 5.1 system, for example -- must, will tend to reduce the ability of the system to recreate the acoustic environment fully.
  
• The Ambisonic "activists" Sweeney refers to are in fact a group of some of the world's leading digital audio experts, recording engineers, producers and audio manufacturers. Their group is called Acoustic Renaissance in Audio and if readers want to check out their proposals for themselves, they are on the Internet at http://www.meridian.co.uk/ara/. It will quickly become clear to any intelligent reader that far from being ravings of fanatics, the modest proposals for the addition of two digital flags in the DVD audio data stream make a lot of sense.
  
  The first and most important ARA proposal is that DVD should include a flag to indicate the use of lossless compression (or "packing"). The systems currently proposed for multi-channel audio on DVD utilize "lossy" compression. The difference is obvious. Compression enables more data to be recorded on the disc. "Lossless" compression schemes are those in which no data are lost in the compression process: everything that goes in one end comes out the other. "Lossy" schemes such as those used by AC-3 and DTS are those in which compression is achieved by removing data which are deemed "less noticeable" -- according, we hope, to some acceptable psychoacoustic principle. Now if you are of the opinion that something sounds the same whether all the data is there or not, consider this example from Dave Malham, Ambisonic expert at the University of York, in England:
  Ths sntnce is readbl. This sentence is readable. But which do you prefer?

The second proposal from the ARA is that there should be a flag to indicate the presence of hierarchically-encoded surround sound systems. A hierarchical surround system is one which offers compatibility all the way from full surround down to mono. The UHJ transmission system utilized by Ambisonics is an example of such a system, though by no means the only one. Ambisonic recordings usually consist of up to four sum-and-difference channels: mono (the sum of all the signals); left minus right, front minus back and, interestingly, up minus down. Notice that this system encodes full with-height surround into four channels: in Ambisonic parlance, it is called "B-Format". Sum and difference signals, however, are not compatible with conventional stereo and mono. As a result, for compatibility purposes, B-Format can be encoded (I didn't say "matrixed") into up to four channels. Depending on the number of channels available for transmission, a set of signals can be derived by the listener's equipment which makes the best use of the available transmission channels. If there are four channels available, the listener can experience with-height surround. Three channels provide high-resolution planar (conventional horizontal) surround. Two channels can be decoded to deliver a respectable planar surround effect, or without decoding provide a "wider than the speakers" super-stereo. And if only one channel is available, the resulting mono is a good deal more representative of the overall content of the recording than summing the channels of a 5.1 recording would be.

I must say that I find the idea of adding two little flags to the DVD data stream: one to allow for future expansion into with-height surround, the other to allow compression techniques to be used which do not degrade the audio quality -- both adding to the possibilities of the format, while taking away nothing -- as a fair way short of heretical. It certainly does not amount to "enshrining the system as part of the... standards for an audio-only DVD". The ARA does not even insist that an audio-only DVD is necessary.

Finally, I do feel that it is worth pointing out the simple fact that to date there have been more recordings created for original album release in Ambisonics than in any other surround-sound format. This is because most other existing surround-sound formats were designed originally for cinematic applications, where impressiveness is a great deal more important than localizational accuracy. As any engineer who has mixed in Dolby Surround will tell you, such systems are really difficult to work with for music mixing. Positioning is very unstable, particularly when sources are moving (static sources wander about too), and in any so-called "discrete" system the listener position is extremely important, while the speaker layout is specifically limited to a square with the "sweet spot" in the center. Ambisonics, on the other hand, does not have a sweet spot (you can enjoy the effect even outside the speaker array) and speakers can be placed wherever is convenient, within reason.

The answer to the question, "If Ambisonics is so wonderful, why isn't everybody using it," is not a technical one. Ambisonics, thanks to being owned for much of its life by a Thatcherite British Government quasi-non-governmental organization which didn't believe in funding, let alone marketing, has received consistent technical praise from some of the most accomplish audio minds in the world, but very little promotion. Its originators are some of the most respected names in digital audio -- such as the late Michael Gerzon. But as any former Betamax owner will tell you, technical prowess seldom stands up to marketing muscle and vast Dolby-sized promotional budgets. Despite this, Ambisonic recordings are still being produced on a regular basis by some of the most respected names in the industry. To choose a just a few random examples: the entire Nimbus Records catalog, for example, is Ambisonic (and I am sure they do not regard themselves as merely "maintaining" it as Sweeney suggests). Paul McCartney's Liverpool Oratorio is Ambisonic. Alan Parsons' Stereotomy is Ambisonic. And Ambisonic techniques were employed on Tina Turner's Break Every Rule. And that's just the start.

Ambisonics has an entirely respectable track record, and is more than keeping pace with today's emerging multi-channel digital audio transmission standards. In a world where standards and systems are continually developing, it is important to see a good breadth of coverage of different systems in the press: but for goodness' sake, please let's have the truth, not biased misrepresentation. You and your magazine can do better than that -- and your readers deserve it.

For more information on the subject, see the author's Ambisonics-related Web pages, The Ambisonic Index, which contain several articles on the subject and links to the majority of international resources on Ambisonics, including the extensive discography maintained by Mark Anderson.

*Richard Elen is a recording engineer and producer with over two hundred albums to his credit, including about fifty recorded in surround-sound of one sort or another. He was the studio manager and chief engineer at EMI Music Publishing's London studios in the Seventies, founder of Sound International magazine and Editor of the leading international professional audio journal Studio Sound from 1980 to 1984. He is now the Vice President of Marketing at Apogee Electronics Corporation, the Santa Monica, California-based manufacturer of high-end A/D and D/A converters. He is has been a Member of the Audio Engineering Society for nearly two decades and lectures regularly on digital audio topics and techniques, and on the history of the recording industry.