Articles and whitepapers

1/6/2003

Surround Sound Formats - Part Two - The Surround Sound Merry-Go-Round: Pipelines, Presenters, and Processors

By Mark F. Davis

The path by which an original surround recording reaches the ultimate listener may be a circuitous one indeed. For example, a cinema sound effect might start out as a stereo field recording, upmixed to 5 channels in post production by a first processor while being mixed with other sounds, conveyed on DVD via a discrete multichannel coder, downmixed to stereo by the DVD player (a second processor), upmixed back to 5 or more channels by a home matrix decoder (a third processor), and either presented via a full-range or satellite speaker system, or downmixed back to binaural with a (fourth) headphone processor and presented via headphones.

With that dizzying array of mix-and-match options, it is understandably difficult to speak about any single 'surround format' in isolation, as the number of possible combinations is already quite large, and growing. So we will try to organize the surround landscape into pipelines, processors, and presenters, before taking a brief look at what might be waiting in the wings.

Pipelines

Pipelines in this context are transmission or storage media used to convey audio from one place to another. Mostly they carry one or more discrete channels of audio.

These days the most common numbers of channels in a pipeline are 1, 2, and 5.1 (6, really). The only single-channel pipelines around are used largely in instances where squeezing in more channels simply is not feasible: AM radio, low bitrate Internet audio and, of course, the telephone. Common stereo pipelines in active use include the Compact Disc, the Sony Mini-disc, much Internet audio, most MP3s (and many other low bitrate coders), the audio tracks on VCRs, FM radio, satellite radio, analogue optical film tracks, TV audio, cassettes, and even the venerable vinyl LP record. The most recent widely-used pipeline format, 5.1 channels, is found on DVDs, Audio-DVDs, digital television, and digital movie soundtracks (the latter also includes the Sony 7.1 channel SDDS system).

Examples of different pipelines bringing
4-channel Dolby surround into the home

Depending on the signals conveyed, each of these pipelines can be considered a format unto itself. Just connect speakers or headphones to the outputs, sit back and enjoy.

One theme that runs through this ensemble of pipelines is the steadily increasing use of signal processing to extract the best possible performance from a given medium. The earlier systems, such as LPs and mono FM radio, made do with just pre- and post-equalization. FM-stereo added sum/difference matrixing for backward compatibility with mono receivers. Cassettes and stereo optical soundtracks became viable hi-fi media when noise reduction was added, principally the Dolby-B system for cassettes and Dolby-A and SR noise reduction for films.

The process essentially started again with digital media. The audio Compact Disc uses raw digital PCM data, while Mini-disc, MP3s, satellite radio, TV digital audio, Internet audio, DVDs, and digital cinema soundtracks all make use of audio coders to reduce the required data rate to a more manageable level. To some extent, the characteristics of those pipelines reflect the characteristics of the respective coders involved.

One of the interesting trends in some of the more recent pipeline formats is the use of 'metadata.' This is optional information that is specified by the content producer and can be used to control possible post-processors. For example, the Dolby Digital format used on DVDs and in some digital TV systems includes metadata specifying how the channels should be weighted if they are downmixed, and a dialogue normalisation level to assure similar playback levels of different programs. It is expected that the continued proliferation of surround formats will be accompanied by increased use of metadata to help deal with the 'mix-and-match' issue.

Presenters

Strictly speaking, there are but two types of widely used presenters: headphones and loudspeakers. There is not much that needs to be said about headphones in isolation. The question of whether or not they can be considered a surround format unto themselves is probably a matter of personal preference.

The world of loudspeaker-based presentation is, however, considerably more complex. For starters, we have the basic arrangements resulting from connecting one of the three mainstream pipelines to speakers: mono, stereo, and 5.1 surround (5 full-range speakers plus a subwoofer). But then there are several different basic types of loudspeaker, including horn, direct-radiator, bi-directional, and phased array. This raises questions about which type is best for a particular installation, and whether different types should be intermixed in a 5.1-channel array. The answer often depends on personal taste, the characteristics of the listening room, and perhaps the preferred types of material to be played over the system. Movie theatres often use horn radiators for the front speakers, in part to maximise the clarity of the dialogue, and arrays of direct radiators for the surround channels, to create a more diffuse surround field. Home theatre systems are more likely to use direct radiators all around. The THX consumer system, developed by Tom Holman, specifies the use of dipole speakers for the surround channels to try to enhance the sense of envelopment.

Even within a given type of loudspeaker, there may be format-related options, particularly with regard to the reproduction of low frequencies. A 5.1-channel system using 5 full-range speakers might sound best when the bass is fed to all of the speakers. A similar system using small satellite speakers should preferably feed all low frequency energy to the subwoofer. Of course, a system with full range speakers but no subwoofer will probably sound best with the bass distributed among all the speakers, while a system using full-range speakers for the Left front and Right front channels, but satellites elsewhere, should feed the bass to the two full-range speakers, possibly augmented by a subwoofer. And if stereo material is played over a system with a subwoofer, it is generally necessary to have a crossover filter to separate the bass for the subwoofer from the signals going to the other speakers. Given this array of possibilities, it is often desirable for the receiver or amplifiers in use to include some sort of bass-management circuitry, to allow selection of bass redirection to best match the corresponding speaker array.

Processors

A processor in this context takes some number of channels as input, perhaps with metadata or other information, and generates a number of output channels that may or may not be the same as the number of input channels. Over the course of time, an enormous number of such systems has been devised, and discussion here is necessarily limited to those in current widespread use. Many of these systems are strongly associated with a particular type of pipeline or speaker array, such as 2 to N upmixers and N to 2 downmixers.

Examples of encoding, upmixing and downmixing

Downmixers, which reduce the number of channels, are probably the simplest of these, usually consisting of passive signal addition networks or software routines. These processors are ubiquitous, inexpensive, and are often built into equipment which may be required to put out fewer channels than available at the input, such as the aforementioned DVD players that provide a stereo output from a 5.1 channel soundtrack. Passive downmixers are prone to two problems: overload from adding high level signals together, and a failure to maintain total signal power when correlated signals are downmixed. These problems are usually manageable with simple means for existing formats, but may become more of an issue, requiring active processors, if future systems allow for more channels.

Since downmixers generate output channels by forming weighted sums of the input channels, they are sometimes referred to as 'matrix' systems, since similar operations govern the cascade of mathematical matrices. Actually, most upmixers, which increase the number of channels, are matrix systems as well. Early matrix upmixers appeared during the age of Quadraphonic sound in the 1970s, and were pioneered by Peter Scheiber. Their purpose was to try to recover four channels from a stereo pair. Initially, such systems were also passive, but this resulted in unacceptably low interchannel separation, so a variety of active systems were introduced to enhance the apparent separation. The Dolby Surround system used on 'Star Wars' and subsequent pictures had to use an active two-to-four matrix system, because there was not enough room on the film for four discrete optical channels.

The large and growing installed base of 5.1 channel speaker systems together with the continued popularity of stereo content has spawned a modern-day revival of matrix processors to upmix two channels to five or more. Representative algorithms include the Dolby Pro Logic II system, designed by veteran matrix engineer James Fosgate, the DTS Neo:6 processor, which can steer signals at different frequencies independently, and the Lexicon Logic7 system, designed by David Griesinger, which produces seven wideband channels from two, to provide both side and rear surround channel pairs.

These and other matrix systems can use either conventional stereo material or specially encoded two-channel content derived by downmixing multichannel sources. Their highly individual characteristics and the many operational modes supported by each mean that on any particular program, one or another system may provide a more pleasing presentation, depending on one's taste. So receiver manufacturers have taken to including multiple processors in some receivers, leaving the ultimate choice to the consumer.

In recent years, the standard discrete 5.1 channel arrangement has been extended by combining it with a matrix decoder to extract an additional centre rear channel. This system, known as Surround EX, was developed collaboratively by Lucasfilm and Dolby.

One matrix system that stands apart from the others is 'Ambisonics,' invented by noted British engineer Michael Gerzon. Rather than trying to create isolated images at the speakers, Ambisonics makes near-constant use of all the available speakers to create an approximation to the original sound field in the vicinity of the listener. The effect can be quite compelling. Ambisonics is unusual in being able to work with a variable number of source channels and speakers, with four channels and eight speakers representing a typical configuration. Unfortunately, the system tends to have a limited 'sweet spot,' ill-suited for audience use, and it can be difficult to transcode to and from other surround formats.

Another popular surround processor of a different stripe is the acoustic crossfeed canceller, originally developed by Shroeder and Atal in 1963. This system prepares signals for a pair of speakers such as to allow virtual images outside the span of the speakers, and can therefore be used to create a virtual surround system with just two speakers. This system is found on PCs, in some TV sets, and standalone home theatre systems, but usually has a limited sweet spot, and so is most appropriate for a single listener.

Lest we forget the world of headphone-based listening, there are sophisticated processors that impart localization cues to a stereo or downmixed 5.1-channel program to create the perception of a full surround presentation when auditioned with headphones. The best known of these is the Dolby Headphone system, developed by Australian R&D firm Lake Technology.

Looking Ahead

With such a vast array of surround sound formats and processes available to consumers, it might be supposed that audio engineers could sit back and relax, but such is not the case. In laboratories around the world, there are some intriguing candidates for next-generation formats being developed.

The coming adoption of digital cinema in place of 35mm film is likely to bring with it an increased number of audio channels available to film soundtrack mixers, which should afford some attractive creative possibilities.

One possible arrangement is the '10.2' system proposed by Tom Holman, which adds to the standard 5.1 channel array extra speakers along the sides of the theatre and above the screen, plus a second subwoofer.

There are also companies working on ultrasonic speakers that project beams of sound into a room, allowing sound to appear to come from any location without the need for other loudspeakers. A small British firm, 1-Limited, recently perfected an audio-band beam speaker using 253 drivers which projects five separate beams to create a full surround presentation. This system is being marketed by Pioneer.

And there is an advanced system using panels of phased-array loudspeakers called 'Project Carrouso' being developed by a consortium of European companies. This system may allow faithful simulation of a full 3D sound field in an auditorium.

Summing Up

By now it should be clear that the age of digital audio has spawned a cornucopia of surround formats available to the consumer. This is a bit of a double-edged sword, as the blessings of having lots of choices are balanced by the need to actually go out and listen to some of these systems if one is to make an informed choice.

The options are further complicated by the expansion of viable venues and 'digital convergence.' No longer are we limited to hearing surround sound in cinemas or home theatres; there are automotive DVD players and surround audio systems of considerable sophistication to while away the hours on the open road. Once out of the car, there are recently introduced 'Walkman-style' portable video players (and, sometime soon, video mobile phones) that will let you watch movies while out hiking. And helping to tie it all together are home entertainment networks and servers to allow access to your personal media collection from anywhere in the house.

By all accounts, we seem to have collectively arrived at what might be fairly called 'The First Golden Age of Surround Sound.' Enjoy.

Mark F. Davis is the Senior Engineer at Dolby Laboratories, Inc.

www.dolby.com