Whatever Happened to Ambisonics?

So What Happened?

The NRDC approach consisted of obtaining and administering the patentsassociated with an invention, funding its development, and then finding alicensee for the invention. The inventors would then earn a royalty from theinvention and the NRDC would recoup its investment. That was the theory, and itworked quite well in some areas. The NRDC approach would particularly suit youif you had, for example, developed a new way of making some kind of plastic.The idea could be licensed exclusively to a chemical manufacturer, and off yougo. For some other types of invention, however, the NRDC approach was disastrous--astheinventor of the Hovercraft would testify.

In hindsight, some might propose that the NRDC was not the best organisation tohandle Ambisonics. However, although a lot of things didn't happen, and anumber of apparently ill-advised things did, it is difficult to see what otherorganisation would have got Ambisonics off the ground. The system may havelanguished for a decade or so, but it is quite possible that without the NRDCit wouldn't be here at all.

The problem was that while the NRDC was set up perfectly to license aninvention to one exclusive licensee--ideal with that chemical process, forexample--it was not in the slightest bit in a position to promote a systemwhose success rested on as many companies as possible becoming licensees. Wecan imagine that it would have failed equally with an invention like CompactDisc, DAT, Dolby B, or even the humble Compact Cassette. While the NRDC had thefunding to go around selling ideas to individual companies, the idea of massmarketing an invention like Ambisonics --holding big press conferences,exhibiting at trade shows, making demonstration records, and generally sellingone thing to a lot of people-- was out of the question. Ambisonics neededsomething more like product marketing and less like searching quietly for anexclusive licensee. It is even possible that the NRDC's brief simply didn'tallow it to do the things that Ambisonics needed.

One by one, however, companies began to pick up on the system: record companieslike Nimbus--the longest and most consistent licensee, with literally hundredsof CDs produced over the last 25 years, every one Ambisonically recorded withtheir equivalent of a Soundfield microphone; hi-fi manufacturers; andprofessional audio manufacturers like Calrec, who produced the first Soundfieldmicrophone.

Ambisonics was originally designed to reproduce sonic actuality as accuratelyas possible--an approach exemplified by Nimbus Records, whose 'Natural Sound'is more an entire philosophy than simply a method ofrecording and playback: surround sound handled as accurately as possible isjust one of the facets of the Nimbus approach. However, there were plenty ofpeople who wanted to do decidedly unnatural things with the process, likemixing multitrack recordings Ambisonically.

The idea of Ambisonic panpots had been included in the original theoreticalwork by Michael Gerzon, but it wasn't until the early 1980s that practicalpieces of studio equipment began to emerge--from Audio & DesignRecording--which enabled conventional multitrack recordings to be mixed into anAmbisonic format. These units (now licensed to Cepiar) were--and stillare--very cost-effective, and several major artists began to use them, but inthe meantime, apart from Nimbus, very little was happening.

A chicken-and-egg situation developed during the Seventies and early Eightiesin which hardware manufacturers looked at Ambisonics but put their projects onhold due to lack of software--they were looking for more than a series ofclassical CDs, however good they were. Boots Audio, for example, were poised tolaunch a complete Ambisonic microsystem-- but changed their minds. Meanwhile,many people on the record side were unwilling to make Ambisonic recordingsbecause nobody could decode them.

This situation should never have arisen, and it could have been short-circuitedby two things, had they been better known. The first was thatAmbisonically-recorded albums sounded a lot better than regular stereo, even ifyou didn't have a decoder. For example, Digital Audio magazine in 1986reviewedone of the first mainstream Ambisonically-mixed CDs-- Stereotomy, byAlan Parsons--with comments like, "Studio pop production doesn't get anybetter... a winner in the sound quality stakes. Sounds emerge from everywhere,clear and clean. The opening of track 3... completely fooled my dog intothinking a car had driven up the driveway. The only track in which Ambisonicswas not used... [was at] lower volume, more distant." It was worth makingAmbisonic records, even if nobody ever decoded them. And secondly, asmanufacturers like Minim and Troy Ambisonic (a maker of in-car Ambisonicsystems) quickly discovered, decoders could offer a 'super stereo' mode whichwould dramatically enhance existing stereo recordings played through thedecoder, by extracting surround information and using it to create impressivelocalisation and 'wrap around' effects.

And the strategy of trying to persuade record companies to endorse Ambisonicsand use it on all their albums--a similar approach to that used by the failedQuad systems--really put the companies off. Anything that smacks of doubleinventory is likely to do that. And besides, not only was there no need to askrecord companies to commit to Ambisonics; you didn't need to ask them at all,any more than if you wanted to use a particular make of digital reverb on youralbum. The decision was made in the studio by the producer, not by someone atthe record company. It was only very late in the day that direct approachesbegan to be made to producers and studio personnel, and then NRDC fell foul ofthe next problem -- the government that created it.

It's a known fact that Margaret Thatcher's government really didn't like the idea oftheNRDC. Their view was that British inventions should stand or fall on theirability to attract industry backing on their own, and that a "quango" -- a quasinon-governmental organisation -- shouldn'tdo it for them. But rather than admit this, the course taken was to restrictthe NRDC and prevent it doing its job properly, so as to demonstrate how suchorganisations were a Bad Thing--a technique which was also attempted withthe British Health Service. The NRDC was bound together with the National EnterpriseBoard--who at the time used most of their budget to fund British Leyland--toform a fictitious entity called the 'British Technology Group'.

Not too long after this, despite highly competent NRDC people in charge of theAmbisonic project--as was the case all along, it is important to point out--virtually everything that was being done, stopped being done. At the time, amember of staff privately suggested to me that one of the main problems wasthat nobody knew how much funding they'd have next month, so the idea ofplanning anything like a long-term marketing plan for Ambisonics was completelyout of the question. The system became moribund, with a few exceptions: NimbusRecords; a few other enterprising record companies like Brendan Hearne's YorkAmbisonic; parts of the BBC quietly doing drama and concert recordings withSoundfield mics; EMI Music's KPM Production Music Library; and manufacturerslike Calrec, Audio & Design, and Minim.

Eventually, the NRDC saw a way out of the situation, simply by doing what theywere best at--namely locating a single, exclusive licensee and letting themtake responsibility for 'doing something' with the mass of Ambisonictechnology, which by now included nearly 400 patents. Very soon there werethree contenders for the privilege: Nimbus Records, Avesco plc, and a Canadiangroup called Maple Technology. To the likes of you and me, Nimbus, with twodecades' experience of the system, were number one contenders; and Avesco, amajor British technology-based group with interests in high technology audioand video, were second. None of us knew anything about Maple, so it was a greatsurprise when they were awarded the licence. Then everything went quietagain--for months. Absolutely nothing happened and eventually the licence wasterminated. Next, Avesco got it--and also proceeded to sit on the technologyfor months. They disposed of their Troy Ambisonic subsidiary (a condition ofobtaining the licence, apparently!). After a long period of inactivity, theytoo lost the licence.

Finally, the exclusive licence to Ambisonics passed to Nimbus Records. One ofthe first activities of then Company Secretary Stuart Garman--an avid musicenthusiast and long-term supporter of Ambisonics--was to present the system tomajor Japanese manufacturers looking for a way to offer new, serious surroundsound capabilities in their products. A UHJ decoder will handle DolbySurround-encoded material very impressively, interestingly enough, (it is alsopossible to convert Dolby Surround material to UHJ) and UHJ itself is an idealaudio format for future TV and disc formats and Digital Audio Broadcasting. And abuilt-in 'super stereo' processing mode ensures that any stereo recording willsound impressive, UHJ encoded or not.

First to pick up the technology was Mitsubishi, for their 'Home TheatreSystem', a fully-integrated audio/video component series. The decoder in theMitsubishi DA-P7000 system was implemented entirely in the digital domain, thefirst commercial product of its kind. Since then, at least two other majorhi-fi manufacturers (Onkyo and Meridian) have gone into development with thesystem and more announcements are expected shortly. [These units are apparentlystill in their respective catalogues -- RE]

Meanwhile, on the software front, Collins Classics, at the time an increasinglyimportantclassical label, announced their intention to record all their albumsAmbisonically--not in the Nimbus way with a single microphone array, but usingmultitrack digital recorders and Ambisonic mixing equipment. The announcementfollowed a series of experiments with the technology, including recordings ofVaughan Williams symphonies with Sir Neville Marriner. Ambisonics, at last, wasgetting the attention it deserved.

Using Ambisonics

Natural Sound Recording

Ambisonic Mixing

Because of the nature of the transcoding process, the Transcoder cannotgenerate B-Format. It is also difficult to pan around the room, as the consolepanpots are limited to panning across the front or rear stages. If dynamiceffects are required, the Pan-Rotate unit can be used. This takes a mono input(typically from post-fade channel out on the console via the patchbay) andallows it to be positioned anywhere in the planar Ambisonic soundfield. Acontinuously rotating panpot sets up the direction of the sound, while another"radius vector" control enables the apparent distance of the sound from thecentre to be varied, from full positive, through zero at the centre, to fullnegative (ie. panning across a diameter of the soundfield). Each Pan-Rotateunit will handle up to eight mono inputs. There is a rotate control whichrotates the entire signal generated by the unit. Additional B-Format inputs cancome in either before or after the master rotate control, so that units can bedaisychained, and the B-Format output is usually fed to a Transcoder's B-Formatinput for UHJ encoding. Typically, the Pan-Rotate unit is used for mix elementswhich need to be moved during the course of the mix, while the Transcoder isused for elements which remain in their positions. [It can also be used totransfer 5.1 recordings into B-Format --RE.]

A more accurate, but seldom used, unit is the B-Format Converter. This enablesstandard console panpots to be used to generate B-Format signals. These canthen be fed into a B-Format input on a Pan-Rotate unit or straight into aTranscoder. The Converter is designed to operate with constant-power consolepanpots but it operates entirely satisfactorily with the compromise betweenconstant-power and constant-voltage now generally found on mixing consoles. Anauxiliary send is derived postfade and set to the same level as the faderoutput. This can be done by measurement or by ear and provides the 'W' (mono)component of the B-Format signal. Then four groups are fed into the unit.Selecting a pair of these groups (one odd, one even, as is standard practice)and panning between them allows panning across one quadrant (90 degrees) of the360 degree Ambisonic soundstage. This unit is generally used in combinationwith the other two.

Whatever mixing method you use, experience indicates that if pan positions aredetermined while listening in surround, you can then return to stereomonitoring and concentrate on that for the rest of the mix: the surround willtake care of itself. In fact, in general, it is advisable to create the finalbalance while monitoring with the most basic configuration the material is likely tobe heard on. In other words, if most of your listeners will hear the materialin mono, monitor in mono as you do your final balance. If most people will hearit in stereo, monitor in stereo. The resulting surround balance will be fine.

Stereo Enhancement

Ambisonics was designed originally as an encode/decode system, in the same wayas Dolby, and it is undeniable that Ambisonic recordings are best experiencedvia a decoder with a multi-speaker system. A typical stereo listening setup would beexpected to have speakers at 60 degrees for the front stage, and in itssimplest form an Ambisonic decoder just adds two speakers to the rear in thesame configuration. When decoded, a horizontal surround Ambisonic system can beused to localise a source anywhere within a circle. Every position in thecircle is represented by a unique combination of phase and level, once again.In fact, in Ambisonics, the phase/level combinations have been psychoacoustically setupto closely emulate the relationships actually experienced in our hearing.

When the decoder is bypassed (or missing), the rear signals are 'folded over'to the front speakers This means that something at the far left or right edgesof the circle will actually fall outside the speakers in stereo, when thedecoder is switched out and only the standard stereo speakers are used. Signalsin the rear soundstage additionally have a more distant quality--when undecodedtheir level is reduced slightly to enhance this effect--and generally appearbehind the listener due to 'aural decoding' (the brain attempts to localise thesound source correctly, based on the phase and level information).

Ambisonics can therefore rightly be considered as a 'stereo spatial enhancementsystem'. In fact, the results of listening to 'undecoded' Ambisonic mixes arevery impressive. Mono compatibility is excellent and there are no 'forbiddenpositions' which don't work in mono.

The spatial effects experienced with nondecoded Ambisonics are at least asimpressive as those achieved with some of the systems currently in vogue, butthey are more stable and are very independent of listener position--you can getthe effect almost anywhere in the room. There is little or no sound changecaused by different spatial positions and there is no appreciable 'phasiness'.The reproduction of spatial positions is very accurate up to about 180degrees--beyond that strong positions are heard, but they are not as accurateas when the signal is decoded. However, the undecoded results are highlysatisfactory and rival those of other systems--but without the expense or theshortcomings.

The mono/stereo compatibility of Ambisonics means that a recording can be madewhich at the same time offers mono listeners an exceptionally accurate balance;stereo listeners a stereo that is much wider and more stable than conventionalpanpotted-mono, and is less dependent on listener position; while listenerswith a decoder can experience a uniquely accurate full surround sound. In aworld in which few people currently have decoders, this 'future-proof' aspectof an Ambisonic recording, offering full compatibility with existing systems,impressive stereo enhancement effects, and with the ability for those samerecordings to be decoded at a later date into full surround, is veryimportant.

Simple stereo enhancement can be carried out by processing existing mixesthrough a Transcoder. This technique is used by several AM stereo radiostations in the USA to make their stereo signal sound more impressive. Simplyfeed the stereo signal into the front stage inputs of the Transcoder and setthe front stage width to maximum. The UHJ output can be treated as stereo. Youcan try adding reverberation, with the reverb returns being brought back to therear stage inputs.

However, it is generally much more impressive to use the Transcoder asdescribed for basic Ambisonic mixing, feeding groups into the unit and mixingwith it rather than processing the balance afterwards. In this case you areactually creating an Ambisonic recording, and if it isn't too much trouble,it's worth listening to it via a decoder and four speakers from time to time.

A Word About Mono

Go to Part I of this article.

For more on Ambisonics, see The Ambisonic Index at www.ambisonic.net.