Report - August, 1995
By Bill Barnes
There has been much debate in some hi-fi publications regarding the sound quality of AC-3. Some people find it difficult to believe that a system which uses relatively few bits (compared to current industry standards such as 16 bit PCM as used in the compact disk) can sound good. In reality, what is important is not just the number of bits that are used; it's how they are used. That's where Dolby has focused their expertise and understanding of psychoacoustic principles. When thinking about the sound quality of AC-3, there are certain facts that should be kept in mind:
a) The standard that all digital audio systems are currently compared to, that is, the "CD quality" sound of 16 bit PCM, is really a very wasteful system. It takes a sort of "brute force" approach which requires little intelligence in either the encoder or decoder, and it uses a large number of bits all the time, even if they are not all needed to accurately code the audio signal. It is certainly possible to design more intelligent (efficient) coding systems which can represent audio signals with greater accuracy than 16 bit PCM while using fewer bits.
b) AC-3 can operate at a variety of data rates, and generally, the higher the data rate, the better the sound quality. So it makes little sense to simply comment on the sound quality of AC-3 in general, without specifying a bit rate. Depending on the bit rate used, just about any coding system can either have obvious artifacts or be audibly transparent.
c) Comparing coding systems running at different data rates is like the proverbial comparison of apples with oranges. In order for a fair comparison to be made, the coders must be running at identical or very similar data rates.
d) The intelligence of AC-3 is built into the encoder, which means that, as more testing is done on the system, the encoder can be made better and better, with the result that AC-3 soundtracks will sound better and better. As we learn how to make the encoder "smarter", so that less coding artifacts are produced, this improvement will be passed on to listeners when the soundtrack is played back. You can think of an AC-3 coder as a computer running a program with many variables, and as better values are discovered for the variables, the program can be easily updated for improved performance. This ensures that the technology can continue to evolve for many years to come. It also means that improvements in sound quality will be automatically passed on to end users with no need to purchase new decoders. In fact, we have already made notable improvements to our encoders, so that new AC-3 laser discs will have higher sound quality than those released just a few months ago. We anticipate that we will be able to continue making improvements for quite some time.
So, with all of that in mind, just what do audiophiles think of the sound of AC-3? Recently, several audio magazine writers and editors had the opportunity to compare the AC-3 laser disc soundtrack for the movie "The Mask" with the Dolby Surround encoded PCM version and with a specially prepared DTS version. (DTS stands for Digital Theater Systems, a company that manufactures digital soundtrack players for movie theaters.) DTS has been mainly active in the digital theater sound market, and there are now several movies which offer DTS soundtracks, along with theaters equipped to play them back. They have proposed a move into the consumer market by releasing laser discs with a DTS soundtrack instead of an AC-3 soundtrack. Because the DTS coding system is, in our opinion, less efficient than AC-3, and thus requires much more data than AC-3 for similar sound quality, DTS proposed including their soundtrack in place of the existing stereo PCM laser disc tracks. Unfortunately, this proposal would leave those without DTS decoders with only the analog FM audio tracks. (It should be noted that one problem with this approach is that the PCM digital tracks are required as part of the laser disc standard, while the analog FM tracks are not.) There had been claims that DTS was more suitable as an audiophile format because it used a much higher data rate than AC-3, and the recent listening test was an opportunity to put such claims to the test.
The listening test involved Thomas Norton and Peter Mitchell, both from Stereophile, David Ranada from Stereo Review, and Gary Reber from Widescreen Review. The soundtracks compared were a six channel discrete master recorded on a Tascam DA-88 eight track digital machine (16 bit PCM), the AC-3 coded version on laser disc (384 kbps data rate, or 384,000 bits per second), a DTS version on laser disc (1.44 Mbps data rate, or 1,440,000 bits per second), and the Dolby Surround encoded two channel PCM soundtrack on laser disc (16 bit PCM). In the end, after hours of critical listening, there was no significant difference heard between the AC-3 and DTS coded soundtracks even though the DTS soundtrack was using approximately four times more data than the AC-3 soundtrack. In addition, there were only minor differences heard between the AC-3 soundtrack and the original discrete digital master. Keep in mind that this represented one of the first AC-3 laser disc releases, and that notable improvements have already been made in the AC-3 coder, as discussed above. As expected, the AC-3 soundtrack represented a significant improvement over the Dolby Surround version played back through a Pro Logic decoder, with the enhanced channel separation and wide bandwidth stereo surround channels providing a much more realistic and involving experience. To sum up, here's what those involved had to say about the sound quality of AC-3 after the test:
Thomas Norton, Stereophile, April 1995
"Regardless of these considerations, I
was very impressed by what I heard from this comparison. So was
everyone else who heard it, it seemed."
"I began the day at Widescreen Review anticipating my eventual in-home assessment of the new Dolby AC-3 discs with real concern, and ended it eagerly looking forward to the experience."
"Even Gary Reber, who has been more critical than anyone of Dolby's reluctance to support before-and-after comparisons of AC-3 coding by high-end press and manufacturers, appeared to be impressed. After the DTS and AC-3 playback equipment was packed up and carted away, he stared at his copies of True Lies and Clear and Present Danger, lamenting the "hidden tracks" that he couldn't play again--at least not until he can obtain a production decoder."
Peter Mitchell, Stereophile, April 1995
"A final note: We ended our day at Widescreen Review by comparing AC-3 playback to the conventional matrixed (Pro Logic-decoded) PCM soundtrack. I've heard several such comparisons in theaters, so it came as no surprise that the AC-3 playback was much more involving and satisfying than the matrixed Pro Logic sound. In some respects--more precise tracking of on-screen and off-screen images, specific left-and right-side localization of split surrounds, and the dramatically increased deep-bass impact provided by the 0.1 channel--the superiority of 5.1 sound was equally evident in a big theater and in the living room. In terms of the subtler aspects of the sound (greater width of the front soundstage, increased clarity of small details, and superior reproduction of realistic ambience), the advantage of discrete 5.1 over matrixed Pro Logic was more dramatically evident in the living room than it has been in theatrical demonstrations. If you're serious about Home Theater, add an AC-3 decoder to your shopping list."
Gary Reber, Widescreen Review, Issue 12
"I personally found the Dolby AC-3 to
perform exceedingly well, in fact, better than expected given the
fact that it was processing data using a much more aggressive
data-reduction or compression algorithm at 384 kilobits per
second for the 5.1 channels. It was virtually transparent to the
printmaster dub except for a tinge brighter character and less
inner detail, all of which were subtle differences. Through much
of the play I heard no differences."
"I did not detect quantization noise or any other artifacts while listening to the soundtrack excerpt."
David Ranada, Stereo Review, June 1995
"Use any of the standard movie-review
cliches--stupendous, exciting, gripping, thrilling, stunning,
marvelous, powerful, spine-tingling, a triumph, two thumbs up,
****(highest rating) --and you'll be understating my first
reaction to what will undoubtedly become a significant audio
technology over the next few years: Dolby Surround AC-3. If you
think Dolby Surround Pro Logic, to use its official name,
provides an involving home theater experience, you ain't heard
"Meanwhile, run, don't walk, to your nearest hi-fi showroom and experience AC-3 in action. If you hear one new audio technology this year, it must be Dolby Surround AC-3."
And finally, a few comments from yet another audio writer who did not participate in this particular listening test, but who heard AC-3 prior to the test:
Corey Greenberg, Home Theater Technology, February 1995
"I'm telling you, nothing in the past 5.1 years has excited me as much about the future of Hi-Fi as Dolby AC-3, and you're going to be just as blown away as I was when you get to hear it."
Part of the appeal of AC-3 goes beyond simply delivering 5.1 channels of audio to the listener. The primary emphasis of the system has always been on providing the highest quality audio with a limited number of bits, but it was realized early on in the development of the system that it could also provide certain features that listeners would find useful. In addition, it has become apparent over the years that there are certain problems inherent in reproducing soundtracks in the home, and AC-3 provided an opportunity to address some of those problems.
For example, the typical wide dynamic range of a movie is not always appropriate for listening in the home. Movies are mixed for playback in the theater at a volume level that is relatively high when compared to the typical playback level in the home. In addition, loud sounds such as explosions are usually much higher in level than average dialog levels. This creates a problem that many home theater owners are probably already aware of, if the volume is set so that the dialog is at a good level, the loud passages will "blow you out of your seat" as I have often heard it described. Alternatively, if the volume is adjusted so that the loud passages are at a comfortable level, the dialog becomes unintelligible. In some movies this can even happen with the musical passages between scenes, and on more than one occasion this has caused me to repeatedly turn the volume down during those passages and then turn it back up again in order to hear the dialog. Doing this throughout a movie definitely detracts from viewer enjoyment!
This wide dynamic range problem is one of the reasons that most programs broadcast on television are compressed. In spite of the recent flurry of interest in large screen televisions and home theater, there are still many viewers with older sets which have small speakers and relatively little audio power. These systems simply cannot reproduce wide dynamic range soundtracks satisfactorily. Because of this, most broadcast soundtracks are intentionally mixed with a restricted dynamic range, and additional compression is usually applied to the soundtrack before broadcast. This practice is designed to produce satisfactory results on "least common denominator" equipment, but unfortunately, the viewer with high quality equipment is left with no way of reversing the process.
This scenario begs the question, wouldn't it be nice to be able to tailor the dynamic range of the soundtrack so that you could listen to it the way you wanted? And furthermore, if the dynamic range was compressed for broadcast, wouldn't it be nice to be able to undo that compression so that you could enjoy the full dynamic impact of the soundtrack in your high performance home theater? Well, AC-3 allows you to do just that. Included within the system is a variable dynamic range function which allows the soundtrack to be compressed to suit the requirements of the listening environment. In addition, there is a provision for a "preferred" compression setting that can be determined by the producer of the program. This means that when a movie is mixed, the full dynamic range soundtrack can be used for the theatrical release, and the producer can choose an appropriate compression setting for other uses, such as home video or broadcast. The same principle applies to soundtracks produced only for broadcast, such as television shows. The producer will be able to mix a wide dynamic range soundtrack and then an appropriate compression setting can be determined for broadcast. When the soundtrack is reproduced at home, the viewer can select the full dynamic range version, the "producer approved" compressed version, or select the amount of compression that suits their taste and listening environment.
Another problem that is no doubt familiar to many television viewers is that of differing audio levels in different programs. As I "channel surf" through the channels on my cable system I inevitably come across a few programs that require a change in volume for comfortable listening. Of course, the volume must then be readjusted when I move on to another program. Likewise, when selecting different program sources such as laser disc, VHS tape, or television broadcast, there can be significant differences in the volume levels of the various sources. To eliminate this problem AC-3 contains a "dialog normalization" feature which is designed to match the average dialog levels of various programs, essentially correcting for variations in subjective loudness between different programs and sources. This level matching will be based upon dialog reference levels, so that, for example, the announcer presenting the evening news can be heard at the same level as actors speaking in a movie, even though the movie will no doubt contain higher peak sound levels. This feature, in combination with the dynamic range control described above, will eliminate much of the need to manually make volume adjustments to compensate for program characteristics.
In addition to these level and dynamic range controls, AC-3 provides a few other useful features. One is the system's flexibility in handling bass frequencies. Even though the system will deliver five full range channels plus a low frequency effects (LFE) channel, it will not be necessary to use five full range speakers plus a subwoofer. Home theater enthusiasts who want the ultimate system will no doubt opt for such a setup, but the more common arrangement will use speakers that are very similar to those currently used for a Pro Logic system. Indeed, for some years to come many users will be playing back a mixture of both regular Dolby Surround and AC-3 soundtracks, so a single speaker arrangement will need to work well for both formats. For this reason, most AC-3 surround products will offer a minimum of three different speaker arrangements, and redirect the bass signals from the various channels accordingly. (I'm using the phrase "AC-3 surround products" to specify those products which offer surround channel outputs, as opposed to products which only offer two channel stereo outputs.) The first option is to use five full range speakers plus a subwoofer. In this mode there is no redirection of bass, and each speaker simply reproduces the full signal contained in the corresponding channel. A second option would be to use a speaker setup that is common in Pro Logic systems, with large left and right front speakers, a smaller center speaker and small surround speakers. In this mode the bass would be removed from the center channel and both surround channels, combined with the LFE channel, and distributed to the front left and right speakers. A third option would be to use five smaller satellite speakers and a subwoofer. In this mode the bass would be removed from all five main channels, combined with the LFE channel, and sent to the subwoofer. This mode will be particularly attractive to many users, because smaller satellite speakers are generally easier to place in typical living rooms, and it will allow the use of identical speakers in the five main channels.
Another useful feature of AC-3 will be the ability to "downmix" multichannel programs. This means that programs with 5.1 channels of information can be mixed down to a smaller number of channels, such as two channels for ordinary stereo listening, or even mono. One particularly attractive aspect of this feature is that an AC-3 surround soundtrack can be downmixed to a two channel Dolby Surround encoded soundtrack. (An "AC-3 surround soundtrack" is one which contains audio in the surround channels, as opposed to a two channel stereo program.) This downmixing ability has a number of advantages. For example, let's say that you have a Dolby Pro Logic Surround system and you are not yet ready to upgrade to an A/V receiver with AC-3 decoding. However, your loca l cable company is offering a new digital cable service that you would like to install, and it delivers only AC-3 audio. You need not worry about incompatibility because the cable set-top box will offer the usual left and right audio outputs, and these outputs will carry a Dolby Surround encoded downmix of the AC-3 soundtrack. They can therefore be connected to your Pro Logic decoder, allowing you to enjoy the programs in Dolby Surround just as you always have. In addition, the set-top box will offer an AC-3 digital output which can be connected to an AC-3 decoder once you decide to upgrade your system. This will also be true for digital satellite systems which use AC-3 audio. One additional benefit that results from the ability to downmix is that even after you have upgraded your system to include an AC-3 decoder you will be able to record the downmixed two channel soundtrack on your existing equipment, such as a VHS recorder. When the program is subsequently played back there will of course be no way to retrieve the original AC-3 bitstream (you would need a recorder with digital recording and playback capability for that, and those are still a year or so away), but you will be able to enjoy Dolby Surround playback through your Pro Logic decoder.
Another feature of AC-3 is the ability to positively identify a soundtrack's original production format, such as mono, regular stereo, regular Dolby Surround, or AC-3 surround. When considering what this might mean to you as a user, it is helpful to think of AC-3 as a "pipeline" which delivers up to 5.1 channels of audio. The pipeline can carry from one to 5.1 channels, and it may carry two channel soundtracks which are Dolby Surround encoded. A/V receivers with AC-3 surround decoding will also include Pro Logic decoding, so they should be able to detect when the incoming AC-3 bitstream is carrying a Dolby Surround soundtrack (in just the left and right channels, of course) and switch to the appropriate decoding mode. By identifying the format of the soundtrack, AC-3 will allow this to happen automatically. When a regular AC-3 bitstream is received, the decoder will simply perform AC-3 decoding, but when an AC-3 bitstream carrying a Dolby Surround soundtrack is received, the decoder will first perform AC-3 decoding to extract the two channels of audio, and then it will perform Pro Logic decoding. All of this can be accomplished by processing the bitstream, entirely in the digital domain, and the change from regular AC-3 playback to Pro Logic playback can be transparent to the listener.
Dolby S-type Noise Reduction
And now for something completely different, a few words about Dolby S-type noise reduction (NR). S-type is Dolby's latest noise reduction system for analog tape recording and playback. It is derived from our professional noise reduction system, SR (for Spectral Recording), which is widely used in studio recording and film production. S-type provides 24 dB of noise reduction at high frequencies and 10 dB of noise reduction at low frequencies, making tape noise inaudible at typical listening levels. In additional, S-type has been designed to provide good compatibility with B-type NR, especially in the types of environments where cassettes are primarily used, such as in automobiles.
To test the sound quality and compatibility of S-type, Dolby arranged a series of three demonstrations at the Winter Consumer Electronics Show (WCES) in January of this year. The demonstrations were open to all attendees of the show, and a variety of audio industry personnel participated. Each participant was given a card with questions about the demonstrations so that they could indicate their preferences. The first demonstration consisted of a comparison of an S-type cassette to a compact disk of the same material. The S-type cassette was taken from the production line of a major dupicator and was high speed duplicated at 80 times normal speed on ferric tape. There was no special preparation of the cassette, and it was made using the duplicator's normal materials, equipment, and procedures. The results of the demonstration even surprised us. 60% of the listeners indicated that they heard little or no difference between the S-type cassette and the CD, and of those that heard a difference, 15% preferred the CD while 25% preferred the S-type cassette.
The second demonstration simulated the playback of an S-type cassette in an automobile. A videotape, recorded in a car while driving on a freeway, was shown, and a Dolby Surround system was used to create a realistic soundfield around the listeners. A decibel meter was used to set the road noise in the demonstration room to match the noise level that was measured in the car when the video was recorded. Two different cassettes of the same program material, one recorded with B-type NR and one recorded with S-type NR, were then played on a car stereo equipped with B-type NR. 88% of the listeners indicated that they preferred the sound of the S-type cassette, while 12% indicated that they preferred the sound of the B-type cassette.
In the third demonstration, the environment of either the beach or a city street (both were used at various times throughout the demonstrations) was recreated using an appropriate video and the Dolby Surround system. Again, two cassettes of the same program material were used, one recorded with B-type NR and one recorded with S-type NR. These two cassettes were played back on a "boom box" style portable player which contained no noise reduction. 90% of the listeners indicated that they preferred the sound of the S-type cassette, while 10% indicated that they preferred the sound of the B-type cassette.
The results of these demonstrations make a strong case for the quality and practicality of S-type. For those who would like to try out S-type for themselves, there are a variety of decks available from manufacturers such as Aiwa, Denon, Harman/Kardon, Onkyo, Pioneer, Sony, Teac and Yamaha. S-type cassettes are currently being manufactured by Warner Bros. Records and BMG, with over 2,500 titles available and well over 60 million copies produced to date.
Dolby on the Web
We are pleased to announce that Dolby Laboratories can now be found on the World Wide Web at http://www.dolby.com. The site was just recently opened and is still being expanded so there are a few rough edges, but please stop by, have a look, and let us know what you think. We plan to make it the source for information on all of Dolby's technologies, so please feel free to let us know if there is anything that you would like see included.
If you would like to read more about any of Dolby's technologies, there are a variety of publications available. To request information on any of our technologies, you can send an e-mail message to firstname.lastname@example.org or leave a voice message on Dolby's Literature Hotline at 415-558-0344. Please be sure to include your name and mailing address in your message and specify which technologies you would like to receive information on. You may also retrieve most of Dolby's literature from the Dolby Audio/Video Forum on America Online (the keyword is "Dolby"), or visit our newly opened Web site at http://www.dolby.com.
Dolby Licensing Laboratories
San Francisco, California
Other related articles
© Copyright 1995, 1996, 1997 Secrets of Home Theater & High Fidelity
Return to Table of Contents for this Issue.