Multichannel Film Sound
By Michael Karagosian
© 1999 MKPE Consulting All rights reserved worldwide
Published in Film Journal Dec 1999 and S&VC Jan 2000
Revised September, 2003 with Elisabeth Lochen
© 2003 MKPE Consulting LLC All rights reserved worldwide
In The Beginning…
Multichannel sound in the cinema has its roots in the 1939 release of Disney's Fantasia. Fantasound, as it was called, used three optically recorded discrete sound tracks - Left, Center, and Right - and a fourth control track used for synchronization. The format used clever means to route audio to as many as 100 speakers all around the auditorium. The complexity of the system and World War II contributed to the demise of the system, and Fantasia was the sole release using this format. It wasn't until the 1950's when a process for placing magnetic stripes on both 35mm and 70mm film was developed, and magnetically recorded multichannel sound-on-film was born. During that time, the Cinemascope sound format from Twentieth Century Fox was introduced, which placed four channels of magnetically recorded audio on film Left, Center, Right, Surround -- and Todd-AO introduced the 6-track format for 70mm films Left, Left Extra, Center Right Extra, Right, Surround.
No improvements were made to that arrangement until the 1970's, when Dolby Laboratories introduced two new concepts to film sound. One was the implementation of noise reduction in both magnetic and optical tracks. The second was the introduction of quadraphonic matrixed audio to stereo optical recordings. By adding noise reduction to magnetic tracks, Dolby increased the dynamic range of the medium, and was able to enforce a standard flux level for on-film recording, further enhancing their ability to control the quality of magnetic recordings. After making a few movies with noise reduction added to the Todd-AO 70mm format, Dolby decided to use the LE, RE speakers as extended bass boxes only, and not as full bandwidth sound tracks. Thus, the discrete Subwoofer track was born, often referred to today as the LFE track (low frequency effects). This move caused the 70mm sound format to be effectively reduced to 4-channels plus subwoofer. In 1979, Dolby improved the 70mm format by adding stereo surround with the release of Apocalypse Now. The stereo surround process placed the upper frequencies of the stereo surround channels on the LE, RE channels, and added the bass signal from the mono Surround track to form full bandwidth stereo surround signals. This format enjoyed few releases, however, as the popularity of 70mm film waned. It was, however, the first step towards the 5.1 format now popular today, having Left, Center, Right, Left Surround Right Surround, and Subwoofer channels. Some large installations added the full-bandwidth mono surround to the rear of the theatre, which was an early step towards 3-channel surround. (Michael Todd of Todd-AO also described a 3-channel surround system in the 1950's.)
As high in quality as it can be, magnetic recording on film was, and continues to be, expensive. For this reason, Dolby's other process, matrixed audio for stereo optical recordings, was heralded as a major breakthrough in low-cost, multichannel audio for film. In the studio, a matrix encoder reduces four discrete channels to two. In the cinema, a matrix decoder derives Left, Center, Right, and Surround signals from the stereo optical sound track. Matrixed audio has several drawbacks, however. The format only supports a mono surround signal, and the matrix process itself isn't very true to the original discrete 4-track recording. In its struggle to get it right, Dolby implemented several matrix encode/decode processes in the early years of the format. They settled on a combination of technologies called variable matrix decoding and equal sound power. In the variable matrix scheme, a different decoding matrix is switched into the listening path based on the dominance of either Left, Center, Right, or Surround signals. Each decoding matrix is designed to maintain sound power in the listening environment, so when it is "steering", the matrix decoder has the least perceptible effect on the total energy in the sound field.
Creating four tracks from two leaves a lot to be desired, however. When the Center channel is dominant (a condition that occurs whenever dialog is present), the Left and Right channels become monophonic. The same occurs whenever Surround is dominant. Panning, or any placement of sound other than a hard pan to a speaker, becomes very tricky. To work around these flaws, a special monitoring device is used that allows the mixers to actually hear their mix as it will be heard in the cinema while they are mixing. Film mixers who have worked extensively with the process will tell you of the many tricks that they have to employ to obtain normal sounding results when mixing for the matrix.
In the late 1970's, Dolby Laboratories introduced another element into the film world that largely goes unrecognized, and that is the advent of the stand-alone cinema sound processor. To make the new sophisticated formats simple to install, it became necessary for Dolby to sell a single product that contained stereo optical photo-cell preamplification, noise reduction, matrix decoder, equalization, format selection, and a single volume control for all channels. Thus the cinema processor came into existence. To accommodate new formats along with legacy formats, the cinema processor concept continues to be the mainstay of the industry. Today the manufacturing base has widened considerably, and we also find quality cinema processor products on the market from Sony, Ultra Stereo, and Smart Devices, among others.
And Along Comes Digital Audio
In 1990, the Optical Radiation Corporation in collaboration with the Eastman Kodak Company introduced the world's first digital audio format for film. Called Cinema Digital Sound, or CDS, it was first introduced to the public with the 70mm release of Dick Tracy. Unfortunately, it only enjoyed a handful of releases before ORC pulled the format from the marketplace. CDS was a highly engineered technology, incorporating an error detection and correction scheme in the printed bit map to ensure robustness of the digital signal. The bit map itself was impressive. Thanks to the support of Eastman Kodak and specially developed film stock, the format boasted a very tiny pixel size, giving the format lots of digital bandwidth on the film. Audio was recorded without data compression, but instead of the commonly used 16-bit linear PCM scale, it used a custom 12-bit logarithmic scale that produced excellent sound with a 90 dB dynamic range. Despite its clever engineering, CDS suffered from a major shortcoming, that being that the actual recorded digital sound track for 35mm films was placed where the stereo optical track normally resides. In other words, there wasn't a stereo optical track that could be used as a backup. CDS also suffered from playback equipment problems, which, combined with the absence of a backup track, turned the major filmmakers away from the format.
FIGURE 1. CDS Sound Track
Optical Radiation's CDS format should be remembered as the pioneer for today's digital film sound formats. They were the first to introduce digital audio to the theatre, and the first to implement a discrete version of today's popular 5.1 technique in a major exhibition format. The impact of the 5.1 format has been tremendous. From an artistic point-of-view, discrete digital sound has been a real boon, as it frees the mixers from the limitations and hassle of mixing audio for the matrix. Full range and discrete surround channels have given the mixers the ability to use these channels much more freely. If one were to listen to the mixes of today and those of 10 years ago, the degree of artistic freedom that the digital formats offer will be obvious.
More formats were soon to follow. In late 1991, L.C. Concept became the first system to commercially employ digital sound-on-disk. The soundtrack was placed on 2 x 5.25-inch rewritable magneto-optical disks manufactured by Sony. The 2 disks were capable of storing 300MB of data each, providing up to 3 hours of multichannel audio. The 4 or 5.1 discrete channels were compressed with the MUSICAM-algorithm (Masking Pattern Universal Subband Integrated Coding and Multiplexing). A standard SMPTE time-code track was recorded alongside the analog soundtrack, by which the disks were synchronized. The L.C. Concept format, developed by French Elisabeth Lochen and Pascal Chedeville, was first demonstrated in 1991 with a experimental re-release of the 1990 film Cyrano de Bergerac. The first commercial release was in late 1991 with Bis ans Ende der Welt, only in France. In 1994 the system was modified for use with CD-ROMs instead of magneto-optical disks, but it was never put to use commercially. The system had no technical drawbacks (unlike CDS), but the L.C. Concept company failed in 1994 due to a lack of funding. In total, about 30 films were released with L.C. Concept soundtracks, including Cliffhanger, Basic Instinct, Between Heaven and Earth, Bitter Moon, and The Lover.
Although it was announced in late 1990, it wasn't until 1992 when Dolby Laboratories first introduced the SR/D format, known today as Dolby Digital, with the release of Batman Returns. The development that made this format possible was the AC3 audio data compression algorithm for 5.1 audio signals. Not wanting to repeat the mistakes of CDS, Dolby took pains to preserve the analog stereo optical recording format on the film alongside the new digital recording. Real estate on film is precious, and so they opted to record the bit map between the sprocket holes. A precarious place for sensitive digital data, no doubt, but there it remains. Dolby also chose a much larger pixel size than CDS (approximately 1.1 mil x 1.1 mil), and chose to use a 48KHz sample rate. Thus, the digital bandwidth available for SR/D was much lower than for CDS, and consequently AC3 delivers a highly compressed audio signal. The format was first named SR/D presumably to emphasize the fact that stereo optical using SR noise reduction remained on the film. Dolby's format does not employ error correction, and so the format relies on the matrixed stereo optical track as backup when a digital decoding failures occur which can happen with film wear or even splices.
More digital formats appeared in the cinema marketplace. Digital Theater Systems became the 2nd company employ digital sound-on-disk, by introducing the DTS digital 5.1 format in 1993 with the release of Jurassic Park. DTS took a novel departure from both CDS and Dolby by placing audio on CDROMs in a proprietary format and recording only an analog synchronization track on the film. The DTS process has proven to be a very robust method for implementing digital audio on film. In the event of loss of sync, DTS audio will free run for several seconds before switching over to a backup track. Once the time code track is valid, the processor noiselessly switches into the correct audio frame. On the film, the time code track resides alongside the stereo optical track, so as with Dolby Digital, the format has the luxury of an analog backup track. On the CDROM, DTS uses a variation of the APT data compression algorithm, compressing 16 bit audio at a 44.1KHz sample rate. DTS offers a few variations on their format. The commonly used exhibition format records 5-channels of audio, placing the Subwoofer channel on the stereo surround tracks below 80 Hz, and thus rolling up the stereo surrounds at 80Hz. A discretely recorded 6-channel version of the format is commonly used in special venue theatres.
The film industry is known for its competition and innovation, and Sony did not want to limit its productions to a paltry 5.1 format. Thus in 1993 Sony introduced the SDDS digital audio format with the dual releases of In The Line of Fire and Last Action Hero. Unlike Dolby Digital and DTS, SDDS is a 7.1 format, resurrecting the full range LE, RE speakers of the Todd-AO 70mm magnetic format days, but now calling them Left Center and Right Center, and combining them with discrete stereo surrounds and a separate subwoofer track. As with Dolby Digital, the SDDS bit map is recorded directly on the film. Real estate on the film being ever more scare, Sony engineers chose to place their bit maps on both edges of the film, outside of the sprocket holes. As with Dolby Digital, this is also a precarious place for sensitive data. Unlike Dolby Digital, Sony chose to print duplicate bit maps on each edge of the film for redundancy. Thus, SDDS strives hard to insure a valid digital bit stream and is the only format since CDS to offer some method of error correction. As with the other digital formats, SDDS relies upon the stereo optical track for backup. SDDS uses a proprietary data compression scheme, reproducing 16 bit audio data at a 44.1KHz sample rate. SDDS uses a slightly smaller pixel size than Dolby, (approximately .9 mil x .9 mil) and not having to dodge sprocket holes gives them plenty of bandwidth for a healthy not-too-compressed 8 channels of audio.
Not all SDDS releases are recorded in the full 7.1 format, however. Historically, most releases have been recorded in a 5.1 format, and the Left Center and Right Center channels are largely unused. However, it is possible for the Sony SDDS decoder to create artificial LC, RC channels when needed. Conversely, for a film that is recorded in 7.1 format but played on a 5.1 system, the SDDS decoder will fold the LC, RC channels into the L, C, R speakers.
FIGURE 2. Edge of Modern Day Print Showing DTS Time Code, Dolby Digital, and Sony SDDS
New formats continue, and in 1999 Dolby Laboratories introduced an extension to the 5.1 format by adding a center surround signal to the rear assembly of speakers. Called the Surround EX format, there is nothing particularly innovative about the technique, as it uses the good ol' matrix encoder and decoder to place the additional track on the 5.1 stereo surround channel. In fact, carrying the idea further, it is possible to add a "top of auditorium" speaker due to the capability of the matrix decoder. (See Figure 3.) The EX format brings with it all of the hassle and limitations previously experienced with stereo optical tracks, but now with digital stereo surround tracks. While an extra channel certainly adds a new creative element for the film producer, the reintroduction of matrix mixing has not made EX the darling of Hollywood's post-production community. It is interesting to note that Sony's SDDS data format has the capacity for three discrete surround channels. Sony has not explored this as a product, however, as in the words of Sony's Dan Taylor, "we feel that the 7.1 format has yet to be fully exploited".
FIGURE 3. 5.1 System With Matrixed Surrounds
It is important to note that each of the existing digital formats occupies an exclusive area of the film (see Figure 2). In practice, it is possible, and in fact more and more common, to release the film print with the bit maps or time code for more than one format. The question is often asked as to whom will win the "digital format war". This author's answer is everyone. Cinema is a creative industry, and film producers enjoy choice and innovation. There are limitations and advantages to each of the formats, in terms of sonic capabilities, in terms of distribution capability, and in terms of the economics of the film print itself. As long as film is around, there will be a variety of formats for multichannel sound. Nor will electronic cinema alter the multiplicity of formats. Having even more data space to play with, electronic cinema will only increase the number of options available to the film producer.
The earliest cinema processor offerings from Dolby Laboratories were the CP100 70mm/35mm processor and the smaller 35mm only CP50 processor. Over the years, Dolby has released a variety of processors to replace these products, including the CP200, CP65, CP55, CP45, and CP500. The CP200, first introduced in the early 1980's, was the first processor to support stereo surrounds. It has remained the choice of studios for its flexibility, although the unit is no longer in production. Among other brands, one can find today a variety of 5.1 processors, including the Peavey CinemAcoustics CA-CP600, Sony DFP-3000 (also a 7.1 processor), Ultra Stereo JSX-1000, and Smart Devices MODV.
Decoding digital audio from film requires proprietary technology from the company that owns the technology. Dolby, for instance, offers the DA20 decoder and an optical bit map reader, which produce an analog 5.1 signal that can be reproduced by any 5.1 cinema processor. DTS manufactures the DTS-6D, which plays the DTS CDROM discs and includes the optical time code reader that mounts on the projector. The DTS-6D not only has the usual analog 5.1 output, but also a digital output for direct digital connection to a cinema processor. To date, only the CinemAcoustics CA-CP600 digital cinema processor takes advantage of the DTS direct digital audio port. Up until very recently, Sony offered the DFP-D2000 and a corresponding optical bit map reader. The DFP-D2000 did not fit into the traditional cinema processor mold by offering it's own equalization, level control, and bypass relays for peaceful co-existence with a cinema processor. Ultra Stereo, however, offers a clever device that will allow the DFP-D2000 to be used as a source unit to traditional cinema processors. Sony has not announced a stand-alone decoder replacement for this product. The block diagram of Figure 4 show the interconnection that could be required of a cinema processor if each format was installed at the same screen.
FIGURE 4. A Possible 5.1 Cinema Processor Interconnection Showing All Formats
The most recent trend in cinema processors has been to include the proprietary digital processing in the cinema processors made by the companies that own the formats. Dolby Laboratories includes their Dolby Digital decoder in the CP500D. Sony includes their SDDS decoder in the new DFP-3000. DTS offers their DTS-6AD cinema processor incorporating the circuitry of the DTS-6D playback unit. None of these processors can directly support the digital audio format of their competitors, meaning that stand-alone decoders will continue to exist if each format is to gain the widest possible number of installations.
Supporting multiple formats in the film world requires flexible installations in the cinemas. In the cinema's projection booth, it is not uncommon, and in fact standard with many cinema chains, to install "floating decoders." These are digital decoders or playback units that can roll about on carts and plug in to the cinema processor for any screen. In many cases, screens are not installed with a digital system and require a floating decoder if a digital feature is to be shown. As the market saturates, the sale of stand-alone decoders dwindles. But the use of floating decoders in the projection booth remains and is more widespread than most format providers acknowledge.
THX is a well-known but often misunderstood name in the cinema marketplace. It is commonly thought by the general public, and wrongly so, that the Lucasfilm THX program incorporates its own sound format. In fact, the THX program is designed to create a system of standard theatres, all of which meet criteria produced by THX. In pursuit of their standard, THX approves products for use in theatres, which in turn are licensed to use the THX name. THX also involves itself in the design of licensee theatre facilities, observing isolation from neighboring screens, room reverberation, and even HVAC-related noise. THX performs a very real and very useful service to the listening public. But they do not produce a sound format.
Dolby Laboratories first introduced the concept of "A" chain and "B" chain in describing their cinema systems to field technicians. "A" chain incorporates the sound source and related adjustments. "B" chain incorporates the loudspeaker equalization, level, crossover, amplification, and the loudspeakers themselves. Time has proven this to be a useful way of describing the cinema system. One can inject pink noise at a reference level into the "B" chain and setup the listening level and equalization of the theatre. In adjusting the optical sound track playback, for instance, one can make the various level and equalization adjustments required of the "A" chain without affecting the settings for the "B" chain. Keeping the "A" chain and "B" chain adjustments separate in the field technician's mind is important for the correct alignment of a system.
The standard listening level for any full-range loudspeaker in a theatre is 85 dBc SPL. In the case of stereo surrounds, they are treated as a single surround source, and individually adjusted for 82 dBc SPL. This level is measured using pink noise played at the reference RMS level of the cinema processor. When Dolby noise reduction was introduced to the cinema environment, the concept of reference level was easy to promote, as it was simply the "Dolby Level" required of the noise reduction. In the digital environment, "Dolby Level" does not exist. However, -20dBFS (20 dB below full scale) is the point of reference commonly used by the various digital formats.
One of the unique and interesting aspects of cinema sound is the X-curve. "Curve X" is described in ANSI PH22.202M-1984 / SMPTE 202M as a high frequency roll-off for a typical 500 seat theatre of 3 dB per octave beginning at 2KHz, and 6 dB per octave at 10KHz. Correction factors are given for other sizes of theatres something that is largely overlooked in this industry! All cinema playback systems today are aligned to the X-Curve. The history of the X-curve begins with Dolby's early attempts to promote faithful reproduction of sound in theatres with fairly common and not too sophisticated measurement techniques. X-curve resulted from a study of cinema acoustics across the country some 25 years ago. The frequency response described by X-curve was designed to compensate for the reverberation and screen roll-off of the average theatre, giving a uniform static frequency response to which all rooms are to be adjusted.
The problem with X-curve is that it is now 25 years old, and it is questionable that the average theatre room acoustics found today are the same as those when the standard was created. We have lost sight of the fact that the goal is to achieve a flat spectral response in the room after environmental conditions such as room reverberation and air absorption are taken into account. Instead, X-Curve has become an end to itself. The most obvious aberration that it has caused is the Home THX "re-EQ" standard in Home THX-approved receivers, which serves as a translation of X-curve, originally meant for large theatres, to very small room acoustics. The industry would be better served by adopting a suite of curves that represent a variety of room sizes with various reverberation responses, and employing sophisticated analysis and alignment in the studios.
On the subject of theatre alignments, it is worth discussing the 10dB on in-band gain required of Dolby Digital in the Subwoofer channel. The Dolby DA20, Dolby's stand-alone playback product for decoding Dolby Digital films, produces a reference level output for all channels with Dolby's test film, including the Subwoofer channel. Once the Subwoofer response is flattened and the 1/3 octave levels are set to match those of the screen speakers, Dolby requires the Subwoofer level to be increased an additional 10dB. The Subwoofer track will be recorded 10dB lower than otherwise by introducing this boost in the acoustic mixing environment. While the 10dB of extra headroom is not necessary for digital tracks, according to David Gray of Dolby Laboratories, "it's a practice that carried over from 70mm magnetic film track production". Although not so in earlier days, all digital formats now observe the 10dB boost for the Subwoofer track.
Multichannel cinema is rich in variety and innovation. This article has attempted to describe the various systems in use today in an accurate manner. In the never ending quest of the film producer to have interesting and memorable dialog, music, and effects, we can expect multichannel cinema to become richer and ever more varied with time. Electronic cinema will be no exception.
Dolby Digital is the trademark of Dolby Laboratories, Inc.
SDDS is the trademark of the Sony Corporation
DTS is the trademark of Digital Theatre Systems, Inc.