Along the Digital Data Frontier

by Assistant Professor of Music Technology Jeff Baust

Perhaps you've heard the new buzzword "DVD" being tossed around lately, or possibly you've seen a small but growing section of your local video rental store marked DVD Titles and have wondered what they were. Most importantly, perhaps you've heard about the newly announced DVD-audio format and have been wondering just what impact DVD might have on the music industry. Let's take a look at the technology that promises to alter our sonic experience.

	(There is also a newly created DVD-14 format, which combines a DVD-9 layer with a DVD-5 layer, for a storage capacity of 13.2 GBytes.)

First off, DVD (an acronym for Digital Versatile Disc) is nothing more than a delivery format for digital data. Like the venerable compact disc, the DVD holds digital data--information of some kind stored as zeroes and ones. On the CD, the most common types of data stored are audio and computer data. Similarly, the DVD can come in many data formats, including DVD-V (for video), DVD-ROM (for computers), and DVD-A (for audio). The DVD-V discs are the ones you see at your local video store.

The shape of things to come
The basic physical characteristics of a DVD are essentially the same as those of a CD--binary digital information is stored as "pits" created in a single, continuous track that starts at the center of the disc and spirals outward. These pits are read off the disc by a laser and are processed as digital information (audio, video, or computer data). However, on the DVD, the laser uses a shorter wavelength than the CD, is more precise, and therefore can deal with smaller pit sizes and with tracks (consecutive levels of the spiral) that are packed more tightly. This allows approximately seven times more information to be stored on the DVD than on the CD. Figure 1 shows the tiny pits on a CD in which digital information is stored.

Furthermore, the DVD specification allows for the manufacture of DVDs in more than one physical configuration. For example, unlike the CD format, DVDs can have information either on just one side, or on both sides, doubling the amount of information a single disc can hold. Also, a DVD can store data on two "layers" on a single side, doubling the capacity for that side. You can have combinations of single and double layers and sides. The figures below show the basic DVD storage formats by name, their various physical configurations, and disc capacities.

Clearly all of these are capable of holding substantially more information than a CD, which stores approximately 680 Mbytes. For example, the DVD-18, at 17 GBytes, can hold about as much information as 25 CDs!

DVD-Video
While the DVD-V (whether it's a DVD-5, DVD-10, or another capacity) is primarily designed for the delivery of movies, the audio specification of the DVD-V format is also of interest to audio professionals. This audio specification allows for a number of new and interesting sonic possibilities not found on current CDs.

One big feature of DVD-V audio is the inclusion of support for both high-definition digital audio and surround-sound formats. The DVD spec requires that all DVD-V players handle two-channel digital audio in a format called LPCM (Linear Pulse Code Modulation) at sampling rates of either 48k or 96k, and at resolutions of 16, 20, and 24 bits. This digital audio format, LPCM, is the same one used on the audio CD, although the audio CD supports LPCM only with a sampling rate of 44.1k and a resolution of 16 bits. It's interesting, then, that the DVD-V format, which is primarily designed for the playback of movies, is capable of higher quality audio than the current consumer format standard for CDs.

Additionally, the DVD-V can contain multichannel audio for surround-sound listening, arranged in what is called the 5.1 format. Although there are other possible surround-sound configurations (7.1, for example), the 5.1 format is the most widely implemented. It carries five channels of full range audio, feeding those channels to three speakers across the front of the listener and two speakers to the rear, plus one audio channel for low frequency effects (the "1" of "5.1"). Figure 6 illustrates a typical 5.1 surround-sound configuration.

Gently down the streams
Since the data throughput rate (the maximum rate at which data can be sent "down the data pipe") on a DVD-V is 6.144 mbps (megabits per second), this limits the amount of audio information that can be passed through the data pipe. This means that these multichannel formats need to be stored and played back in a data-compressed format. On DVD-V, the de facto standard format for compressing the 5.1 multichannel audio is called Dolby Digital (DD). DD is a lossy compression scheme, which means that when the audio data is compressed, some of it is discarded and can never be recovered, although the compression scheme has been optimized to discard audio information that the listener probably wouldn't perceive. Dolby Digital, like any audio compression format, needs to be decoded as it is played back, by either the DVD-V player or the home stereo receiver.

There are other multichannel audio formats that are optional in the DVD-V spec and can be used either in addition to or instead of Dolby Digital. As with Dolby Digital, they are lossy compression schemes, and the same issues in audio quality apply. In fact, the DVD-V can hold up to eight different "streams" of audio, up to the capacity of the disc, and each stream can be a different audio format (LPCM, DD, etc). For DVD movies, one of these eight audio streams can hold the original audio for the movie, while alternate audio streams might contain director's commentary, alternate language tracks, and so forth.

DVD-audio
In February of 1999, version 1.0 of the DVD-A specification was finalized. This specification allows for a very wide range of digital audio options and offers all kinds of delivery choices for the content provider of audio programs.

For starters, the DVD-A supports LPCM in a wider range of sampling rate options: 44.1k, 48k, 88.2k, 96k, 176.4k, and 192k, at bit resolutions of 16, 20, and 24 bits. At the highest sampling rate and resolution, this yields a theoretical audio performance of 96k bandwidth and 144dB of dynamic range! Note that there are now supported sample rates that are integer multiples of the current CD standard of 44.1k. This makes it easier to record a project at, say, twice that rate (88.2k), and do a simple downsampling to 44.1k for CD release, whereas downsampling from 96k to 44.1k is a more complex process with potentially greater loss in sonic quality. As in the DVD-V format, the DVD-A format supports bit resolutions up to 24 bits. However, unlike the DVD-V spec, the DVD-A has no lossy data compression schemes as a mandatory consideration, although provisions have been made to utilize them as options.

In terms of audio program time, a single layer of a DVD-A can hold more than 74 minutes of two-channel programming at the highest sampling rate (192k) and highest bit resolution (24 bits). If the audio was recorded at 88.2k, 24 bits, you could store around 144 minutes of two-channel audio, and at 44.1k, 20 bits, it could be as much as 318 minutes of program time. In comparison, a CD holds about 74 minutes of two-channel audio at 44.1k, 16 bits.

As with the DVD-V format, there is a data throughput issue that limits the rate of data that can be sent down the data pipe. But at 9.6 mbps, it is much faster than the DVD-V rate, allowing the DVD-A spec to support two-channel audio playback at the higher sample rates (176.4k or 192k), or playback of multichannel (5.1) audio at 44.1k/16 bits without data compression (which would be needed for playback on a DVD-V). However, even with the higher data rate, six-channel (5.1) audio is not capable at the 96k/24 bit format. There is too much data.

To make playback of six channels of 96k/24 bit audio possible, another data compression format has been made a mandatory part of the DVD-A spec: Meridian Lossless Packing (MLP). MLP achieves a data compression of somewhat less than 2:1. It removes redundancy in the audio signal and is thus a program-dependent scheme, so the actual data compression ratio can vary. However, unlike the data compression schemes used in DVD-V, MLP is a lossless scheme, which means that the audio data, when uncompressed, is identical to the original data before compression. Using this scheme, a DVD-A layer could hold between 75 and 140 minutes of 5.1 multichannel audio at 96k/24 bits with no compromise in sonic quality.

Downmixing multichannel audio
What if you don't own a multichannel audio system? No problem. The DVD-A spec contains a feature called SMART (System Managed Audio Resource Technique). It allows the producer of the DVD-A to create downmixing tables, which specify how the DVD-A player should mix together the six channels of audio for playback on a two-channel (stereo) system. This allows both the content provider and the end user to be sure that playback of six-channel audio programming on two-channel systems will be of optimal quality.

Additionally, the DVD-A spec allows for the addition of "value-added" content on the DVD-A, in the form of text, graphics, Web links, and so forth. While it's not yet clear just how these might be used, certainly the possibilities for extensive liner notes, song lyrics, band member bios and photos, Web page links, and video clips might be of interest to the DVD producer and music consumer.

Two recent additions to the DVD-A spec are watermarking and encryption schemes. The encryption scheme allows for the content provider to prevent the unauthorized digital duplication of the DVD-A content. Digital watermarking does not prevent such copying, but through encoded data in the digital stream, it places copyright and similar information into the digital audio signal that can later be read back to prove copyright ownership and permissions. While this watermarking information is embedded in the digital audio signal, it is inaudible even in critical listening environments and will not be distracting to the consumer.

Playback options
How will we play these DVD-As? Currently, there are many DVD-V players on the market, some for as little as $300. While these players can read both DVD-Vs (for movie playback) and standard CDs (for audio playback), they will most likely not be able to play back DVD-As. There may be production of DVD-A-only players, with portable players being the most likely to appear. The buzz in the industry, as indicated by prototype players announced by Pioneer and others, suggests that manufacturers will probably begin producing "universal" players. They will read and play back both DVD-Vs and DVD-As as well as standard CDs, and will offer a fair number of both audio and video output options. Players of this universal type should begin showing up in stores as early as Christmas 1999.

SACD
Of course, no discussion about the future of audio delivery formats would be complete without a mention of the SuperAudioCD (SACD), a competing digital disc format proposal being championed by Sony and Philips (who are also involved in the development of DVD, of course). The SACD format utilizes the same physical construction as a DVD-9 (single side, dual layer), but with a different data format than DVD-As (making them incompatible). The SACD was released in Japan in mid-1999, and will be released in the United States toward the end of this year.

The SACD proposal allows for the same audio program to be stored twice on the same side of a disc. One of the layers would store the audio program in standard CD format (called Red Book CD), and this layer of the SACD would actually be playable on existing CD players. The other layer of audio data would most likely be a multichannel version of the same audio program stored in a digital audio format called Direct Stream Digital (DSD), requiring an SACD player for playback.

DSD is a high-rate (44.1k x 64 or 2.8224 MHz sampling rate), one-bit digital audio system that yields a theoretical performance of 100kHz bandwidth and 120dB dynamic range. For the most part, it is every bit as good as the DVD-A performance. Using its own lossless data compression scheme (Direct Stream Transfer), which yields a 2:1 compression ratio, the DSD layer of an SACD would offer 74 minutes of multichannel (5.1) audio, again without sonic compromise. Further, the SACD proposal, like the DVD-A format, allows for the inclusion of both text and graphics. As of this writing, there is no provision for video. Finally, as with the DVD-A format, Sony and Philips have announced both encryption and watermarking schemes for the SACD. That should make the format attractive to potential content providers.

What makes the SACD proposal interesting is its backwards-compatibility benefits. Since the SACD contains a layer of audio in Red Book CD format, consumers can begin purchasing SACDs immediately and play them back on existing CD players. Later on, after buying an SACD player, consumers can play those same discs and access the multichannel DSD audio layer. Eventually, the consumer or music professional might, for example, have an SACD player in a surround-sound listening room and a CD player in the car, and would be able to enjoy the same SACD in either listening environment.

Another benefit to the SACD is that a recording using the high-rate, one-bit DSD audio can be easily downconverted into any other LPCM format for product release, be it 44.1k/16 bit for CDs today or 192k/24 bit for DVD-As tomorrow. There would be minimal sonic quality compromise in downsampling. With DVD-A, a recording at 192k that might be slated for CD release (44.1k) would require more complex downsampling and potential sonic compromise.

The evolution of multichannel audio
We are starting to see a number of digital audio delivery formats competing for the consumer dollar. How will all this play out, and what does it mean to the artist and audio professional? It's hard to say. At this point, DVD-V hardware and software seem to be screaming off the shelves, and indications are that this is not a short-term trend. Like the CD, introduced in the age of the vinyl LP, the DVD-V format offers the consumer and content creator a big improvement over VHS tape in terms of quality and convenience. As well, DVD-V offers value-added content (director's remarks, etc.), ease of playability (no more rewinding!), and a greater movie viewing experience (digital audio and video, surround sound, FX channel, and more).

Likewise, DVD-A affords the consumer a much more subtle palette of sonic benefits. For example, with two-channel audio, the DVD-A offers the possibility of better sonic quality than the audio CD provided the end user has both the equipment and the listening environment to enjoy it. However, it remains to be seen whether this increased potential in sonic quality will be a selling point for the average music consumer.

So for DVD-A, the attraction might be the added value in the format, most notably multichannel audio. Surround-sound listening, so popular in movie viewing, does indeed offer the music lover a whole different kind of listening experience. Perhaps if 5.1 surround-sound in home theater systems becomes more commonplace, the music consumer may come to expect 5.1 as a standard audio configuration in the way stereo audio is the standard configuration today. Surround-sound music listening environments might eventually become the norm in the home, the car, and elsewhere. In addition to multichannel listening, the possibility of the inclusion of text and graphics can make the purchase and enjoyment of a DVD-A even more appealing to consumers. Finally, through sales of the universal DVD player, the consumer would not be required to purchase yet another piece of audio equipment exclusively for DVD-A, since one universal player can play back DVD-Vs, DVD-As, and CDs.

For the content provider (artist, producer, or engineer), there are many creative and production possibilities offered by the DVD-A format. Multichannel audio has recently become a major point of interest for audio professionals, and many within the industry seem to regard some manner of multichannel audio as the next logical step in audio production and delivery. The current absence of any single delivery standard (or even production and listening environment standards) has not quelled this interest or prevented the production and release of a good number of surround-sound audio discs in various formats. Even today, there are current music productions slated for two-channel release where the recording, mixing, and archiving of those projects is done with an eye toward a possible future release in a high-resolution surround-sound format. Also, should either DVD-A or SACD really take off, there will certainly be a considerable amount of remixing, remastering, and disc-authoring work to be done on existing and new products at postproduction and mastering facilities.

With DVD-A now defined, the advent of universal DVD players a probability, and SACDs a contender as well, it seems that both the music professional and the music consumer will be seeing some new and exciting changes in the ways music is created and enjoyed.