DVD: the Coming Standard

Someday soon there may be a super Compact Disc with 24 bits and 96 kHz sampling rate. But maybe not that soon. The format of the future disc is hotly controversial.

(Reprinted from issue 54 of UHF Magazine. To purchase the issue, click here.)

If you like what you see in these sample articles, get all of UHF by subscribing

 

Do the concepts of high end hi-fi give you a headache? Take 20 minutes for the UHF Hi-Fi Course.

 

Trying to reach UHF? Our contact page tells you how to find us.

Perhaps we shouldn't even call it the "future" super CD, because it's already here, and you can buy it. You may have seen the Chesky ads for its new 24/96 DVD discs. Is the future here today?
     But Chesky and its allies are not the only players in this game, nor are they the most powerful.
     First, let's look at what's at stake. When the CD format was created in the early 80's, digital technology was relatively primitive, and the format was forever frozen at 16/44. What it means is this.
     1) In writing its data, the digital recorder uses numbers with only 16 digits, rounding off as needed. But we can't say "16 decimal places," because digital uses binary numbers -- strictly ones and zeros -- not decimal numbers. At full level, 16 digits is a lot, but as level drops so does the number of digits available to describe the data.
     2) The sampling circuitry "writes down" a description of the music signal only 44,100 times per second. As we have written before, slow sampling results in clearly audible artifacts.
Uncritical listeners, of course, don't notice any of this, and even some magazines claiming to be concerned about hi-fi are perfectly happy with the CD. Most music producers are not. They are all but unanimous in considering that the low-level performance (the last 20 or 30 dB) of 16 bit, 44.1 kHz digital (known as 16/44 in trade jargon) is unusable for music.
     Several companies have, of course, worked to get all they can from 16/44. Various encoding algorithms have minimized the rounding errors, for instance. The most successful of these is JVC's xrcd process. And Pacific Microsonics' HDCD process is a major step forward, making a standard CD behave as though it had more bits than it does. It requires a decoder at the playback end, however.
     All of these may seem like diversions from the real goal: a CD-like disc with a lot more bits and a much higher sampling rate. The development of DVD for video and computers seemed like just the ticket. And various segments of the audio industry set up committees to come up with a standard for the disc of tomorrow. They've been at it for years now.
     There were, as you might expect, conflicting demands. With home theatre spreading relatively widely, consumers might want extra channels for their music too, so the new disc should have five channels. Ah, but the audiophile community wasn't wild about Dolby Digital (AC-3), which discards supposedly inaudible data, and is acknowledged even by Dolby Labs to be less than completely transparent. How to reconcile all of these demands?

The software rebellion
     While the committees were talking, some manufacturers began lobbying for what they considered the best possible standard: 24 bits of data and a sampling rate of 96 kHz. Chesky, notably, began recording its master tapes in that definition, and doing public demos of 24/96 against 16/44. Even inexperienced listeners were enthusiastic, and Chesky led its own group into a rebellion: the committees could talk all they wanted, but in the meantime the consortium would bring out its own 24/96 discs. How? Well, even if DVD Audio players didn't exist yet, video players do, and all of them have 24 bit converters that can read 24/96. Who needs a new standard? Why not go with what's here?
     Which is what all of the new Chesky discs are about. They are genuine 24/96 discs, and any DVD player can play them. How do they sound? Well, the DVD players we've heard are not audiophile models (though that's about to change), but extra data is always good to work with. The people behind 24/96 are betting on this: with real discs that can be played on real equipment, the "official" standards people will be under intense pressure to come up with something that is at least as good.

The official standard
     It's taken years, but the DVD Forum Working Group 4 (WG-4 for short) has come up with a standard, using lossless, not lossy, compression. The compression system is from well-known British audio manufacturer Meridian, and is known as the "Meridian Lossless Packing" system. Potentially, it could blow away 24/96. The reason: it permits (though it doesn't require) a sampling rate as high as 192 kHz! And multiple channels as well.
     In actual fact, the standard is several standards. Depending on the complexity of the music, MLP can permit more than two hours of 96/24 surround sound, with six channels. Alternatively, it can pack over two hours of two-channel 24/192 music. The same standard would allow discs with 64-minute multiple versions of the same music: six-channel 24/96, 5.1-channel Dolby Digital (that's the same lossy compression system used for DVD Video), and a conventional two-channel 16/44.1 version.
     Got that? The future players would keep this straight by reading a "flag" on the disc and adjusting themselves for the format chosen.
     So are we all set for DVD Audio? Not quite. An anti-copying system must still be chosen for the new system. And that's not all...

A major alternative
A wag once said that standards are wonderful things...and that's why we have so many of them. Well, if one DVD standard is good, aren't two standards twice as good? Don't get us started!
     The first alternative is from a pair of heavyweights: Sony and Philips, who are of course the inventors of the CD and participants in the invention of DVD. Would they steer you wrong? Well, you be the judge.
     Their alternative is known as DSD -- Direct Stream Digital -- which they are already offering as a standard for professional recording. The DSD converter uses oversampling and delta sigma modulation to generate a 2.8224 MHz one-bit signal. Yes, it sounds a lot like bitstream, the inferior CD decoding system, but of course it's not the same. In any case, the bit rate is exactly 64 times the CD's 44.1 kHz rate, and its inventors say it is so "analog-like" that conversion can be done with nothing more than a capacitor.
     Sony and Philips are predicting that they'll have DSD players some time in 1999. In the meantime, the "official" committee has tried to head off a potentially destructive format war by providing for an optional DSD version to be included on the mainstream DVD Audio disc.

An upstart alternative
     The other standard is being proposed by DTS, whose video surround sound system is an alternative to Dolby Digital. That system uses lossy compression, but loses less data than Dolby does. DTS says it is working on an alternative lossless standard for DVD Audio, and will go to court to prevent the rest of the industry from adopting a standard that would leave it out in the cold.
     That hasn't exactly made DTS popular with the industry, but then what's one more standard when we have so many already?

HDCD for the new millennium
     Perhaps you've thought of HDCD as merely a transitional technology, giving CD some of the higher resolution of the future standard, and great to have while we wait for the future to finally get here.
     But Pacific Microsonics says HDCD can live on in the new age, with hidden information adding to perceived resolution of DVD Audio just as it does in present-day HDCD.
     And this may startle you: the company opposes the 96 kHz sampling rate. Why? Because professional recordings done at 96 kHz can't be downsampled to the CD's 44.1 kHz without artifacts caused by numerical interpolation. It proposes 88.2 kHz, which converts without artifacts. The small loss, it says, is more than made up by the greater resolution afforded by HDCD encoding and decoding.
     Well...we always said this was going to take a while.

[TABLE OF CONTENTS] [READING ROOM] [HOME PAGE]