MCS Review On-Line Reprints

The Threat of Dolby Surround

by William Sommerwerck

Reprinted from MultiChannelSound, Vol.1, Nos.4/5, October-November 1986.

Visualize a hardy band of settlers, struggling to eke out a living from the barren plains of the Midwest.  The rains are late; the crops are parched; then — heaven be praised — the rains come!  But the downpour becomes a flood that washes away the crops.

Today, audio ambiphiles are similarly blessed and cursed.  We have spent almost a decade struggling to cultivate surround sound.  Now, thanks to the Dolby Surround system used with hundreds of movie sound tracks, the drought of interest in surround sound has become a flood.  Unfortunately, Dolby Surround (which uses Dolby MP matrix encoding) is the worst possible surround system, which makes its success particularly galling to those who fought for something better.  MP's wide acceptance could end any chance of a psychoacoustically-correct and flexible system (such as Ambisonics) being adopted.

In reality, for the limited purpose for which it is designed — for theatrical presentations — Dolby MP works quite well.  In a movie theater, everyone must hear all the dialog, music and sound effects clearly.  Consequently, the design of MP is intended to present a pleasing and reasonable sound experience to listeners located all about a large hall.

First, instead of the conventional rectangular speaker layout, it calls for left, center and right speakers behind the screen and an array of small speakers across the back of the theater.  The center front speaker handles the dialog, so actors' voices stay centered on the screen, regardless of where one sits.  (Music and sound effects can come from any or all of the front speakers.)  The use of multiple rear speakers allows everyone to hear surround effects equally well, while the level from any one speaker is kept low enough that it doesn't 'blast' nearby listeners.

Second, there is only one rear channel, called the Surround track.  This is primarily a cost-cutting measure, though it also assumes listeners won't be much interested in rear directional effects.

Third, since optical soundtracks are rather noisy, the Surround track is rolled off above 7 kHz (in both recording and playback) and Dolby noise reduction is applied to the signal.

Fourth, since many listeners sit nearer the Surround speakers than the front three, the Surround track is delayed so that frontal sounds will always be heard first.  And, as there is some crosstalk from the front channels to Surround, this delay also permits the Haas (precedence) effect to mask the crosstalk, regardless of the listener's position.

Now, while these features address the real problems involved in getting decent surround sound in a large hall, they have nothing to do with home listening.  First, if you're going to use four speakers, it makes more sense to position them as front and rear pairs, to allow any type of directional movement.  Second, home music sources (LP, cassette, CD) are not noisy; there is no need to limit the bandwidth of the rear signals.  Third, said recordings are usually free of phase jitter that causes front/back crosstalk with motion picture film.  So there's no need for delay.  (The delay also makes it impossible to generate side images.)

You may react, 'So What?  Dolby MP is used only for home video.  I don't see the problem.'  The problem is that there is a lot of talk about using Dolby Surround for non-theatrical recordings.  With three other systems that permit a full-bandwidth panorama of sound across the rear (QS, SQ and UHJ), why would anyone in his right mind use MP, with its single, limited-bandwidth rear channel?

The reason is marketing.  Dolby Surround is becoming the de facto surround-sound standard.  (Since it is such a terrible system for home surround sound, this could have been predicted; consumers have an uncanny knack for picking the lowest-quality products.)  The record producer will naturally encode his album with the system that has the most decoders in use.

The situation is worsened by the public's lack of understanding about surround sound.  Consumers are reluctant to buy decoders without the 'Dolby Surround' logo, even though they may work as well as 'officially-sanctioned' decoders and often include features missing on 'authorized' units.  This is not a matter of speculation.  Denon had to add 'true' Dolby circuitry to their decoder before it would sell.

Another problem is Dolby's licensing policy.  To obtain the coveted 'Dolby Surround' logo, a decoder must include Surround channel high-frequency rolloff, Dolby noise reduction and a 20-ms delay.  As we've seen, these features are not only useless for home listening but antithetical to good surround sound.

Dolby MP threatens all we have worked for over the past seven years — universal, high-quality surround sound.  Its general acceptance would be even worse than having no surround-sound system, because MP fills the surround-sound vacuum with the worst possible system, and, by its broad acceptance, makes it the standard.

Something must be done.  Dolby should take the first step, by recognizing that, for home use, MP is an unacceptable system, both technically and esthetically.  (In fairness, this is only true for decoding; Dolby MP encoding can be easily modified to a smooth, great-circle locus, giving it most of the advantages of UHJ encoding.)  Once they acknowledge this, they should deny licenses to anyone wishing to use MP for non-theatrical purposes (including music videos).  Producers would then be forced to use QS, SQ or UHJ, or not encode for surround at all, any of which would be better than MP.1

Maybe this seems like an extreme position.  After all, why should Dolby give up profits in the name of esthetic considerations?  Well, what better reason is there?  Dolby didn't invent MP simply to make money.  They wanted to enhance the pleasures of moving-going.  Why shouldn't they acknowledge that MP is great for movies but terrible for home use, and block the inappropriate use of the technology?

Peter Scheiber has an even better idea.  He's encouraqing Dolby to broaden its definition of 'Dolby Surround', particularly with regard to what constitutes 'correct' Dolby decoding.  The Aphex AVM-8000 (not to mention UHJ decoders) decode MP extremely well, yet none of these units meets Dolby's licensing requirements.  Since they sound good, why don't they qualify for Dolby's support?  Could it be that they aren't playing patent royalties?  (Aha!)

Peter is also developing an encoding system, compatible with both theatrical and home presentation, that does not obsolete existing decoders.  As it stands, MP encoding isn't too bad.  Though soundtracks are usually mixed in a 'four-point' fashion (Left, Center, Right, Surround), MP can also be treated as a great circle locus, with sounds panned smoothly to any position around the listener.

A rationalized MP encoding would permit sounds to be reproduced at exactly the intended positions (even for side sources) in home listening (using a UHJ-type Ambisonic decoder), while theatrical decoding could continue to use the crude, logic-directed decoding appropriate for movies.2  This would be from the same encoding, so everyone's surround-sound needs could be satisfied.

If MP and UHJ were united, it would be the best thing that ever happened to surround sound.  Let's hope Dolby Labs has the good sense to do it.

Peter Scheiber Comments ...

Your proposal of a 'UHJ/MP hybrid' certainly rings a bell.  In 1971, the 'Goliath' CBS and the 'David' Electro-Voice (both, incidentally, soon to become my patent licensees) were the leading commercial purveyors of matrixed surround sound in this country, with their respective 'SQ' (first called 'StereoQuad') and 'Stereo-4' systems.  They had chosen quite different matrix specifications, resulting in a de facto lack of standardization contrary to the interests of the audio industry and the consumer (Which home decoder to buy?).

Enter the Electronic Industries Association (EIA).  The EIA formed a standardization committee for matrixing practice; John Eargle was the chairman and I was a member, together with representatives of the audio industry.

In order to provide needed industry unity, I suggested a new matrix, for mathematical reasons designated 'tetrahedral', which in essence bridged the gap between the different matrices of SQ and Stereo-4, uniting them into a single, compatible hybrid.  At a session of the EIA committee, chairman Eargle proposed my hybrid matrix, noting that '... its performance in normal stereo would be excellent ...'

The proposal promptly died in birth when the CBS representative, the brilliant engineer Benjamin B. Bauer, firmly stated in substance that what CBS already had was perfect, and the industry would just have to accept it, because in no event would it be altered.  End of story of EV/CBS hybrid.  The EV representative had apparently not brought his slingshot to the EIA committee meeting.  The insurmountable barrier was not technical but rather 'brand conflict' or corporate prestige.

Nonetheless, today there are some positive factors on the technical side that were not present in 1972:  First, in SQ, CBS had elected to go with a matrix very different from that of everyone else. A 1974 British Broadcasting Corporation Development report, The Subjective Performance of Various Quadraphonic Matrix Systems, compared the performance of various matrices of the day, including those of CBS's SQ, Sansui's QS and several forerunners of UHJ.  The conclusion, later suppressed in response to CBS wrath, read in part:  "Eargle points out that any great-circle locus can be 'converted' ... into any other by a linear phase-amplitude matrix ...  This indicates that it is possible ... to use, with a given system, material that was originally coded using another.  There is, however, no similar simple conversion to or from the SQ matrix ... as its Scheiber-sphere locus is so irregular in shape."

In common with all these proposed systems, excepting SQ, Dolby's is a 'great-circle' system with reference to the phase/amplitude sphere of directionality (called the 'Scheiber sphere' by the BBC); and, further, it is essentially an 'amplitude-plane' system, representing a straight-forward extension of restricted, frontal stereo space into unrestricted surround space.  (In the amplitude-plane system, a continuous pan around the listening area generally follows the horizontal great circle of the phase/amplitude sphere, and 'in-phase' represents front, while 'antiphase' represents back.)

This brings us to a second positive factor in today's picture, related to the above-described compatibility factor:  Implicit in the great-circle matrix is the ability to convey the locations of sound sources in any direction with respect to the listening area — potentially including not only every direction in the horizontal plane, but also elevation, given appropriate encoding and decoding hardware, by the way.  The point is that the directional and frequency restrictions at present placed by Dolby in the rear are not inherent in their 'great-circle' system, but are merely specified at present by them to govern the system's practice.

Why must this be?  I think I have enough of an inside view of the patent picture to say that the answer involves less cynical motives than the question of patent royalties.  In looking at the motivations of large corporations from the outside, it is natural to think that policy is always made by calculation in cold blood.  It ain't necessarily so.  Corporations are made up of personalities — people who create, and see that their creations are good.  It is common that they want to see these creations perpetuated in their original, 'pure' form, even in contexts that grow to be larger than the context of the original creation.  Being talented people, they are able to muster very persuasive arguments for their position.  These not infrequently take the form of objective technical arguments — which, upon scrutiny, may depend on unstated premises founded in turn on personal taste.

One such issue is the question of whether the system purveyor should include in the system specification technical restrictions representing his view of esthetically correct use of the system — for example, the assumption that surround audio information used with visual program is a kind of 'special effect' not requiring the same fidelity or localization ability as the frontal audio.

My strong opinion is that such esthetic judgments should not be built into the system:  These judgments are properly the province of the creative system user — the individual program producer.  The producer should be able to put sounds anywhere he wants to around the listener, with full fidelity, if he so desires; or he may elect to use the rear more imply as a surround channel for special effects when he decides this is appropriate.  The system should simply 'get out of the way' to allow the producer this freedom of choice.  The producer should not have his esthetic decisions made in advance by the system purveyor.

As you note, I have been encouraging Dolby, over the past months, to broaden the definition of Dolby Surround so as to permit unrestricted, clear placement of sounds in 360 degrees of space, and unrestricted fidelity.  I am not asking them to acknowledge that MP is an unacceptable matrix, since the problems you point out are not inherent in the matrix, but are rather restrictions that Dolby has chosen to place on its use.  What Dolby giveth, Dolby can taketh away — and, without problems of compatibility with existing usage, for the reasons I have suggested; and also because we are not talking about junking the system, but simply removing restrictions that have been specified for use of the system.

Also, as you mention, I have been working on an encoding system compatible with present theater and home playback hardware.  Other than this compatibility, its reason for existence is improved creative flexibility, notably in clear placement of sounds at any desired location in the sides and rear of the environment.  Of course, a disclosure is in the hands of my patent attorney, and I hope to have the time to get together a demonstration in the near future, as 'one pan is worth a thousand words.'

This brings me to perhaps the most important point:  What this is all about is bringing to the audio art the ability to record and reproduce total space, including the reproduction of all sound sources in their proper locations in space.  This ability to reproduce the total space of the sonic event has been properly called 'the last frontier for audio.'

What we don't want is a 'battle of the brands' like that of the 1970s, wherein each commercial promoter focused its efforts on 'proving' that its practice of what is now called 'surround sound' (then called 'quadraphonics', and may be properly described as 'fully spatial sound') was entirely correct, while everyone else's practice was totally wrong.  'MP' and 'UHJ' are commercial designations by two purvevors of matrix surround directional encoding, and not fundamentally divergent technical approaches.  Where they differ functionally is in numerous specific decisions made by their respective protagonists regarding how to use the phase/amplitude sphere of encoded/decoded directionality.  The question is not one of a hybrid between brands — be they designated 'UHJ', 'MP' or whatever.  It is not whether this brand or that is right, but what we want to see the system, by whatever brand name, capable of doing.  I believe that it is a question of the system protagonists providing the ability for the program producer to make the decisions about where he wants to place sounds, and how he wants them to sound.

When you talk about hybridizing UHJ with Dolby's system, I think you are talking about bringing the directional imaginq abilities of UHJ to Dolby Surround.  On this, as I have stated, the unrestricted directional capability of the phase/amplitude sphere of matrix encoding/decoding is just as available to Dolby, if they want to use it, as to the purveyors of UHJ.  But Dolby's clout with movie studios, consumers, etc., is probably not available on the UHJ side.  For this reason, rather than pressing Dolby to recognize that, in your words, 'MP is an unacceptable system,' I would propose that we thank Dolby for making surround sound a reality in the marketplace and press them to acknowledge that the system does have unlimited-directional, unlimited-fidelity capability.

A 'battle of the brands' can benefit only lawyers.  And, there is no sane need for it.  Dolby has done a powerful job of introducing the concept and the reality of surround sound to the new world of 'the marriage of audio and video.'  With significant market demand, this can be expanded — using the same matrix — into the full reproduction of directionality that is inherent in every 'stereo' pair of channels through full use of the phase/amplitude sphere of encoded directionality.

1It's worth noting that, among pop recordings, all recent releases (from Steve Hackett, Michael Stearns and Alan Parsons) have been UHJ-encoded.  This is no doubt due to the availability of UHJ mixing equipment, but is also a result of UHJ's excellent mono/stereo compatibility, mixing flexibility and outstanding surround effects.

2Some theater owners want rear directional effects from MP.  An MP/UHJ hybrid permits this.  Home listeners with UHJ decoders would hear the effect, though theaters would need a special decoder.  Interestingly, current mixing practice is not to pan the Surround track all the way to the rear; this prevents mono cancellation.  Rear left/right panning would position sounds at about the same point on the encoding locus as the single Surround track, so these sounds would still come from the Surround track with existing decoders!

Since the 1960s, Peter Scheiber has been one of the primary innovators of surround-sound theory and encoding techniques, holding many of the basic patents for surround-sound matrixing.  His continuing work to improve matrix decoding and stereo enhancement resulted in the 360° Spatial Decoder in the late '70s.  Most recently he has collaborated with Jim Fosgate in creating the 360° Space Matrix directional processor.

Reprinted from MultiChannelSound, Vol.1, Nos.4/5, October-November 1986.
Copyright © 1986, 2000 by Laurence A. Clifton

Last updated: May 1, 2004

MCS Review On-Line Reprints