© 1995 Dolby Laboratories, Inc.
GROWTH IN CONSUMER APPLICATIONS OF AC-3
TECHNOLOGIES
THE
STARTING POINT
AC-1 (Adaptive Delta Modulation) was
introduced in the middle 1980s, and initially followed a modest growth
curve in professional and some consumer products (notably home decoders
for the B-MAC satellite television service in Australia). In the middle
of 1991 a system for distributing many channels of high quality audio
direct to consumers' homes (now called Music Choice) was introduced in the USA using equipment made by General Instrument
Corporation (GI). There
has been excellent growth in this service with approximately 700,000
systems in place by the end of 1994. A similar service is being launched
in Europe. GI has announced that it will
change from its current AC-1
technology to AC-3 to take advantage of the improvements offered by AC-3.
However, systems using digital bit rate reduction techniques (AC-2
and AC-3) did not follow for several years. Perhaps the most important
step along the way was the recommendation by the technical committee of
the ATSC (the American Advanced Television Systems Committee) that Dolby
AC-3 technology should be used for the United States Advanced Television
System. This recommendation was then endorsed by a 40 to 1 majority by
the full committee and has been fully documented in an ATSC Standards
Document. We have been reporting on this progress regularly to our
licensees.
The results of this decision were dramatic and
surprisingly immediate. No one expects to see High Definition television
in every home in the USA in the very near future or indeed for a long time
to come. Rather, as with the situation in Japan, we expect to see a slow
but gradual growth beginning at the earliest in 1996. The high cost of
the system alone will ensure that this will not be an explosive new
consumer product. However, it became apparent during the development of
the ATSC proposal that there were many other emerging services that would
like to use the same technical systems for very good reasons,
compatibility being of course high on the list. The surprising fact has
been the speed with which the proposals have been adopted and the large
numbers of products which have already been made and sold in the consumer
market by these other systems.
USA (HARDWARE)
The first application of AC-2
technology was the development of a professional satellite link system,
called the Dolby Model 501 encoder and Model 502 decoder. This unit is
configurable with either 48 and 44.1 kHz sample rate versions of the short
an d moderate delay AC-2 algorithms. This unit allowed for transport of 1
or 2 channels via satellite or telephone modems possessing a data rate
capacity of 128-384 kb/s. A very popular application for these units is
the creation of very high quality audio links over T1 or "switched 56"
telephone company lines. The combination of the Model 501, Model 502, and
an ASCEND terminal adapter is also used for ISDN applications, and the
combination is called the Dolby Fax system.
Another early
application of the AC-2 algorithm was its use in the design of a 950 MHz
aural studio to transmitter link system, called the Dolby DSTL. In this
system the use of the short delay version of AC-2 allowed contribution
quality digital audio service to be established in the previously analog
market. This product also employed a 16 kHz sample rate version of the
moderate delay AC-2 to create an additional auxiliary channel service.
Since the 950 MHz band required high spectrum efficiency,
the bit-rate reduction capabilities of AC-2 were very important.
The first major consumer AC-2 system to be unveiled was the PrimeStar
cable system operated by a consortium of cable owners throughout the USA
using equipment manufactured by GI. Plans
for this system were well advanced before the ATSC decisions were
made, and in fact the PrimeStar system was launched using DigiCipher I
for the video and AC-2 for the audio components of this cable service.
AC-2 was chosen because at the time, almost all sources were two-channel,
the implementation was cheap (integrated circuits cost about $6 for a
two-channel decoder), and AC-2 itself can be changed relatively easily in
the future to AC-3 at the time when the video switches to DigiCipher II.
This system has been a major success, with over 400,000 set-top decoders s
old by the end of December 1994.
The first service using AC-3 was
launched in the summer of 1994 by International Cablecasting Technologies
Inc. (ICT). 120 channels of stereo audio programming are being broadcast
via satellite to consumers equipped with a satellite receiving dish and
a decoder box made by ComStream. The service is called "DMX for
Business" and features a hand-held controller which can also provide
information about the music title, album number, and recording company.
For similar reasons to those outlined above, this service is also
two-channel at present. By December 1994, more than 16,000 subscribers
had signed up for the service.
Impressive as these first services
are, they are surpassed by plans for 1995 and beyond. Over twelve
operators, including Tele-Communications Inc.
(TCI) and
ComCast Corporation (reputedly the world's largest and third largest cable
companies) have committed to purchase set-top terminals for a digital
cable system, with General Instrument
supplying over 3.5 million
DigiCipher units with AC-3 technology. Hewlett-Packard (a licensee of GI)
also has orders for an additional 650,000 of these units.
REST OF THE
WORLD (HARDWARE)
While the USA is the first country to
use AC-3 in a large number of products, other parts of the world are also
starting to introduce services. A service similar to DMX for Business was
due to start in Japan early in 1995, in an existing cable system
operated by Cable Audio Network (CAN). While this is similar to the US
service, it differs in that the consumer will not require a receiving
antenna and decoder. Rather, the satellite signals are received by many
ground stations throughout Japan and are decoded at those stations. The
resulting audio is distributed locally via cable to CAN's 500,000
subscribers. All ground and satellite equipment was in place by the end
of 1994.
INTEGRATED CIRCUITS
To
match all of this activity in products there has been equivalent activity
from the integrated circuit manufacturers. Two-channel ICs are available
now from Zoran (ZR38501) for AC-3 and from Motorola for AC-2. Texas
Instruments and Motorola will have
equivalent devices for AC-3 available
in 1995. LSI Logic is developing two devices, one purely for AC-3
decoding and another device (in conjunction with Hewlett-Packard) which
will decode MPEG-2 and DigiCipher II video with AC-3 and Musicam audio. This
IC will be available in mid-1995. In addition, NJRC in Japan is also
developing a multichannel device for 1995.
Of course, many
general-purpose DSP engines can perform the operations needed to decode
both AC-2 and AC-3; the Motorola 56000 series is a typical example.
We are also actively working with several other companies in several
countries on similar IC developments.
MULTICHANNEL
PROGRESS
As can be seen from the previous discussion,
most of the present-day consumer equipment and delivery systems are
two-channel. However, AC-3 was designed originally as a multichannel
system (and not as a single-channel system which then has to be modified
into multichannel - a dangerous and tricky transformation) since much of
the world was already multichannel when AC-3 was developed. Our work in
the film industry for example has been multichannel for at least the past
15 years, and indeed the genesis of AC-3 was in the recognition of the
need for a multichannel digital on-the-film format. With over 300 film
mixes released (March 1995) in this format, playing at over 1,500 movie
theatres all around the world, there has been a great exposure of the
public to true five channel sound from AC-3 sources. It is these films
which will form the start of a library of true multichannel sound.
However, it is clear that for very many years all films have been made
from many elements which have been retained in the film companies' vaults,
and it is a relatively simple matter to go back to these vaults and make a
true multichannel mix from the se original tapes.
In fact, the
first consumer multichannel sources are already on the market. Pioneer
laser disc players with the AC-3 audio format went on sale in Japan in
December 1994, with the first commercial disks released in February 1995.
This standardized deli very system replaces one of the hardly-used AFM
tracks with an AC-3 bit stream (retaining the second AFM track in mono for
those very few owners of older players without the regular PCM tracks).
At the Winter Consumer Electronics Show (Las Vegas, January 1995)
this system was demonstrated; there were also consumer AC-3 multichannel
products displayed by Audio Design Associates, Denon, Enlightened Audio,
Kenwood, Perreaux, Pioneer (in addition to the laser disc players
mentioned above), Runco, and Yamaha.
Another potential large-scale
system is the Toshiba/Time-Warner
digital video disc (DVD) proposal which
has been very much in the world's newspapers recently. This proposal for
a new playback system uses AC-3 for its sound system.
MULTICHANNEL PERCEPTUAL CODING
INTRODUCTION
Dolby AC-3 is a perceptual digital audio coding technique of unprecedented
efficiency, quality, and versatility. It has been providing multichannel
digital sound in cinemas since 1992, and is being launched in two-channel DBS
applications in 1994. Demonstrations of its ability to deliver multichannel
digital surround sound from both tape- and disc-based home video formats
havealready generated excitement within the consumer electronics industry. And
it has been selected to provide digital surround sound with HDTV broadcasts in
the US, which are expected to begin experimentally in 1996.
As its many announced and potential applications attest, Dolby AC-3 is
unusually versatile. Not a single rigid scheme, it is a flexible family of
processes allowing such parameters as bit-rate and number of channels to be
tailored to particular applications. All variations are based on the same
operating principles, however, and have been engineered to ensure compatibility
among formats and adaptability to future needs.
Most significantly, Dolby AC-3 combines high quality sound with remarkable
data rate efficiency. Although, for example, it enables multichannel surround
sound at a lower bit rate than is needed for just one channel on Compact Disc,
the resulting sound quality remains fully consistent with listener
expectations, thanks to Dolby Laboratories' more than 25 years of experience
developing signal processing systems based upon how sound is perceived.
MULTICHANNEL SOUND: PAST, PRESENT, AND FUTURE
Bell Laboratories' famous early experiments with stereo sound in the 1930s
used three channels. The public heard stereo sound for the first time during
the early 1950s in cinemas that used no less than four channels, and as many as
seven. Yet when stereo finally made it into the home a few years later, it used
only two channels, because that was all the phonograph record could
accommodate. This technological limitation, and not listener preference, led to
two-channel stereo becoming the standard for home sound reproduction.
Filmmakers, however, continued to regard four channels left, center, right,
and surround-as the minimum necessary to create a convincing, lifelike sound
field. This was underscored by the widespread adoption by the motion picture
industry in the late 1970s and early 1980s of the 35 mm Dolby Stereo format,
which matrix-encodes four channels onto two optical soundtracks, then recovers
them upon playback by means of a proprietary sound processor.
When Dolby Stereo films are transferred to two-channel video formats, their
four-channel encoding survives intact. Since the early 1980s, Dolby Surround
has made it possible to decode these multiple channels at home-the first real
break with the consumer tradition of two channel stereo. Today millions of
homes are equipped with multichannel Dolby Surround systems; in fact, more
multichannel, surround sound audio receivers are now sold than two channel
stereo receivers. Dolby Surround program material ranges from videos of
theatrical films to TV shows and music-only CDs, helping "home theater" become
the fastest-growing product category in the consumer electronics marketplace.
With the viability of multichannel sound for consumer formats so well
established, the stage has been set for a more advanced technology that
reproduces sound with unparalleled spatial realism, while providing exceptional
adaptability to varying listening conditions and industry needs. That
technology is Dolby AC-3, and, appropriately enough, its first application has
been cinema sound.
HOW DOLBY AC-3 CAME ABOUT
In 1987, the standardization process for a high-definition television
(HDTV) system began formally in the US. For the system's audio, it was first
proposed that four sound channels be matrix-encoded down to a digitally coded
stereo pair for transmission. But by 1990, four or more discrete channels were
considered preferable, to avoid the compromises of audio matrixing. With the
technology then available, however, this would have required at leasttwice the
bit-rate.
It was at this point that Dolby AC-3 was conceived as a multichannel audio
coder operating at little more than the bit-rate required by two independently
coded channels. Previous adaptive-transform coding technologies such as Dolby
AC-2 had already paved the way, but multichannel coding required further
breakthroughs, including a new bit-allocation technique.
DOLBY STEREO DIGITAL: AC-3 IN THE CINEMA
While conceived and later chosen for HDTV in the US, AC-3 was actually
implemented for the cinema first, making it practical to provide multichannel
digital sound with 35 mm prints. In order to retain an analog track so that
these prints could play in any cinema, it was decided to place the new digital
optical track between the sprocket holes, a key factor in defining its maximum
practical bit-rate. It was also well documented that a
HREF="51.gif">5.1-channel format would best satisfy the requirements of
theatrical film presentation. Altogether, these needs dovetailed with the HDTV
requirements that led to AC-3's conception.
Dolby Stereo Digital, the film sound format with AC-3 as its keystone,
debuted in cinemas in June of 1992. Within less than two years, more than 50
feature films had been released in the new format, and nearly 600 cinemas in 27
countries had been equipped for playback of the digital track. This experience
confirmed that prints with both digital and analog tracks could be manufactured
economically, that such prints would play in any cinema, and that the AC-3
coded digital track provided high audio quality with extraordinary resistance
to wear and tear.
Just as important, Dolby Stereo Digital provides a unique springboard for
consumer formats based on AC-3, enabling the accumulation of invaluable
experience in mixing, recording, and distributing multichannel digital audio.
It is also fostering a library of program material immediately available for
consumer release, and has facilitated the development of cost-efficient IC
decoder technology. Dolby AC-3 is the only multichannel perceptual coding
technology with this kind of real-life experience behind it.
DOLBY SURROUND DIGITAL: AC-3 IN THE HOME
Dolby Surround Digital, also based on AC-3 coding, is the consumer
equivalent of Dolby Stereo Digital film sound, forming the final link from
multichannel program producer to home listener. Like the film format, it
provides separate channels for left, right, and center speakers at the front;
two surround speakers at the sides; and a subwoofer at the listener's option.
Unlike analog Dolby Surround with its single band-limited surround channel
(usually played over two speakers), Dolby Surround Digital features two
completely independent surround channels, each offering the same full range
fidelity as the three front channels. As a result, true stereo surround effects
can be achieved for an expanded sense of depth, localization, and overall
realism. And because Dolby Surround Digital maintains complete separation of
the audio channels, it is as suited to music-only recordings and broadcasts as
it is to video formats. Thus it has the potential to open up new worlds of
multichannel sound reproduction.
That isn't all that Dolby Surround Digital can do. While Dolby Stereo
Digital is heard in cinemas with a full complement of loudspeaker channels, a
standardized playback level, and full dynamic range capabilities, home
listening circumstances vary markedly. Therefore, for Dolby Surround Digital
consumer formats, AC-3 has been designed to satisfy many diverse requirements.
At the outset, at least, while some listeners will have multichannel
systems, most will be listening in mono or conventional stereo. Those with
Dolby Surround systems will want a two channel matrix encoded output from their
decoders. Many listeners may prefer a restricted dynamic range, but others will
wish to experience the full dynamic range of the original signal. Techniques to
satisfy these and other needs have been designed in from the beginning:
- Data identifying each program's original production format-mono, stereo,
matrixed or discrete surround-can be sent to eliminate confusion at playback or
reception.
- Program material can be coded when it is originally mixed so that
subjectively constant, dialogue-keyed loudness is maintained as the listener
switches between program sources. No alteration of program dynamics is
involved, only playback volume.
- Decoders can be designed to provide optimum mix downs from multichannel
programming, such as a matrix-encoded two-track mix for analog Dolby Surround
decoding, a conventional stereo mix, or even a mono mix.
- When programs with wide dynamic range, such as movie soundtracks, are
played at low volume, the system can apply appropriate compression to preserve
low-level content. The degree of compression can be made to vary according to
need.
- The listener can program the Dolby Surround Digital decoder to route
non-directional low bass only to those channels in the system which have
widerange speakers or subwoofers.
Dolby Surround Digital offers a dramatic step forward in listener
involvement and excitement. It provides program producers, directors, recording
engineers, and performers unprecedented creative opportunities. And it offers
remarkable media adaptability within a single, far-reaching technological
framework.
DOLBY AC-3 DELIVERY FORMATS
A wide variety of AC-3-based formats is technically feasible, and several
areexpected to make their commercial debut well before the end of the decade.
- HDTV is the first format to announce the use of AC-3 coding. The
Grand Alliance has selected it to provide digital surround sound for the HDTV
system being developed in the US, with the 1996 Olympics targeted for the first
experimental broadcasts.
- Digital cable TV systems are natural candidates for AC-3, both for
its performance and by reason of standardization with HDTV; the first
transmissions are expected by the end of 1994. A two channel version of AC-3
has been developed for such applications, since available programming today is
either mono, stereo, or stereo with matrixed surround. When multichannel
digital programs are eventually transmitted, set-top converters will create a
composite stereo mix down.
- LaserDisc, long the ultimate quality source of movies on video for
home theatre, has been demonstrated in prototype form with Dolby Surround
Digital multichannel sound. Substituting an AC-3 bit stream for one of the AFM
channels leaves the stereo PCM digital audio and a mono AFM track intact, for
compatibility with any existing player.
- DBS and DAB, like cable, are prime candidates for AC-3's unique
combinationof spectrum efficiency, quality, and multichannel capability. The
first DBS system to utilize AC-3 is slated to begin transmitting in 1994 up to
120 stereo music channels via a single transponder for use by business
establishments.
Other applications, including very low bit rate coding suitable for
medicaltelemetry, are under development. Potential users and system
developersshould contact Dolby Laboratories for assistance in applying AC-3
technology.
DOLBY AC-3 IC'S
A major step in the progress of Dolby Surround Digital has been the
introduction of Zoran Corporation's ZR38000, the first single-chip DSP capable
of implementing AC-3 decoding. It is being used first in the second-generation
Dolby Stereo Digital theatre decoder unit.
A lower cost, dedicated AC-3 decoder IC, particularly suited to consumer
products and also providing Pro Logic decoding, is slated for release by Zoran
in mid-1994, as is a two-channel chip for such applications as cable TV set-top
decoders. Other semiconductor manufacturers are expected to begin supplying
AC-3 decoder ICs in late 1994 or early 1995.
ABOUT DOLBY AND DIGITAL
AC-3 is the third perceptual coding system developed by Dolby
Laboratories. Each one combines a unique knowledge of psycho acoustics with a
progression of advanced digital signal processing techniques.
Dolby AC-l was first used in 1985 for DBS applications by the Australian
Broadcasting Corporation. Partly because of its very low decoder cost, AC-l has
since been adopted for other DBS services, satellite communication networks,
and digital "cable radio" systems; the data rate is 220-325 kb/s per channel
depending on application. A refined form of adaptive deltamodulation (ADM), the
data stream contains information not on the absolute value of the audio signal,
but on the change in value from sample to sample. Techniques adapted from Dolby
noise reduction, such as continually varying step-size and pre-emphasis,
greatly improve on basic ADM performance.
Dolby AC-2 uses advanced adaptive transform coding for professional audio
transmission and storage applications; its data rate is 128 or 192 kb/s per
audio channel. Frequency-domain signal processing in a multiplicity of narrow
bands takes full advantage of noise masking, resulting in effective data rate
reduction combined with high signal transparency. Among other applications,
AC-2 is used to link remotely located recording studios and/or film post
production facilities for long-distance, real-time recording, mixing, and ADR
sessions. It is also at the heart of the Dolby DSTL ® system used to link
broadcasters' studios and transmitters.
HOW DOLBY AC-3 WORKS
The digital audio coding used on Compact Discs (16-bit PCM) yields a total
range of 96 dB from the loudest sound to the noise floor. This is achieved by
taking 16-bit samples 44,100 times per second for each channel, an amount of
data often too immense to store or transmit economically, especially when
multiple channels are required. As a result, new forms of digital audio
coding--often known as "perceptual coding"--have been developed to allow the
use of lower data rates with a minimum of perceived degradation of sound
quality.
Dolby AC-3 is the first perceptual coding designed specifically to code
multichannel digital audio. It is also the only one to benefit from the
development of two other successful perceptual coding systems, Dolby AC-1 and
AC-2, and from the development of what are in essence analog perceptual
coding systems: the full gamut of Dolby professional and consumer noise
reduction systems. Indeed, Dolby Laboratories' unique experience with audio
noise reduction is essential to AC-3's effective data rate reduction: the fewer
the bits used to describe an audio signal, the greater the noise.
Dolby noise reduction works by lowering the noise when no audio signal is
present, while allowing strong audio signals to cover or mask the noise at
other times. Thus it takes advantage of the psycho acoustic phenomenon known as
auditory masking. Even when audio signals are present in some parts of
the spectrum, Dolby NR reduces the noise in the other parts so the noise
remains imperceptible. This is because audio signals can only mask noise that
occurs at nearby frequencies.
AC-3 has been designed to take maximum advantage of human auditory masking.
It divides the audio spectrum of each channel into narrow frequency bands of
different sizes optimized with respect to the frequency selectivity of human
hearing. This makes it possible to sharply filter coding noise so that it
isforced to stay very close in frequency to the frequency components of the
audio signal being coded. By reducing or eliminating coding noise wherever
there are no audio signals to mask it, the sound quality of the original signal
can be subjectively preserved. In this key respect, a perceptual coding system
like AC-3 is essentially a form of very selective and powerful noise reduction.
In Dolby AC-3, bits are distributed among the filter bands as needed by the
particular frequency spectrum or dynamic nature of the program. A built-in
model of auditory masking allows the coder to alter its frequency selectivity
(as well as time resolution) to make sure that a sufficient number of bits are
used to describe the audio signal in each band, thus ensuring noise is fully
masked. AC-3 also decides how the bits are distributed among the various
channels from a common bit pool. This technique allows channels with greater
frequency content to demand more data than sparsely occupied channels, for
example, or strong sounds in one channel to provide masking for noise in other
channels.
Dolby AC-3's sophisticated masking model and shared bit pool arrangement
are key factors in its extraordinary spectrum efficiency. Furthermore, where
other coding systems have to use considerable (and precious) data to carry
instructions for their decoders, AC-3 can use proportionally more of the
transmitted data to represent audio, which means higher sound quality.
Technically speaking, AC-3 can process at least 20-bit dynamic range
digital audio signals over a frequency range from 20 Hz to 20kHz x 0.5dB (-3dB
at 3Hz and 20.3 kHz). The bass effects channel covers 20 to 120 Hz x0.5 dB (-3
dB at 3 and 121 Hz). Sampling rates of 32, 44.1, and 48 kHz are supported. Data
rates range from as low as 32 kb/s for a single mono channel to as high as 640
kb/s, thereby covering a wide range of requirements. Typical applications
include 384 kb/s for 5.1-channel Dolby Surround Digital consumer formats, and 192 kb/s for two-channel audio distribution.
[ Up |
Author Index ]