DVD: One Standard to Rule Over Them?
There's still one area of the proposed standard that sucks big-time, however: the multi-channel surround-sound audio soundtracks for DVD's MPEG-2-encoded digital movies will still be squeezed down to a rate of a few hundred kilobits/second using a lossy compression algorithm, most probably Dolby's AC-3. Every piece of Home Theater market research indicates that the improvement in sound quality drives purchases (that and the ability to watch live sports (footnote 1) with enhanced picture quality). Play a movie with better sound, and everyone thinks the picture quality has been improved. Yet the video and cinema industries appear to think it a good idea to level down the sound quality on DVD in order to free up data space for multiple-language soundtracks.
Is lossy audio data compression really that bad? As I write these words, I've just returned from the 99th Audio Engineering Society Convention, held at New York's Jacob Javits Center last October (full report next month). Over and over again, the message at the Convention from engineers who have been evaluating lossy data-reduction algorithms in rigorous listening tests was that not one was audibly transparent! Sure, there were some kinds of music or sound in which listeners couldn't detect any degradation—but every algorithm failed most of the time to pass the transparency test, and every one failed catastrophically on some signals.
In one AES workshop, for instance, we were played one such example: the sound of a pitch pipe. Because of the simple nature of the signal—three tones accompanied by wind noise—there was no psychoacoustic masking, and all the compression artifacts were gloriously audible. Complex film soundtracks are kinder, in that there are plenty of places for the compression nasties to hide, but the worry is that what one sensitive person might hear now, everyone will hear in five years' time.
The irony is, you wouldn't need to steal much bandwidth from the DVD's video data to vastly improve the prospects for audio, perhaps even making possible the use of a lossless algorithm. But am I alone in thinking that for the consumer electronics industry to lock into a standard featuring a lossy audio algorithm with audibly demonstrable flaws is madness?
Regarding the High Quality Audio Disc prospects for DVD, things look more promising. As Peter Mitchell writes in this month's "Update," the most exciting prospect for HQAD is that the one disc would carry every quality option. At its most basic level, a music HQAD would include some kind of data-reduced music signal in order to be compatible with video DVD players. The next step up in quality would be "Red-Book CD Standard" two-channel data, and this would be encoded on just one of the data layers, allowing HQADs to be played back on conventional CD players. The ultimate quality signal—multi-channel, 24-bit-precision, with perhaps a 96kHz sampling rate—would coexist with the other two formats on the same disc, but would need a new player to be accessible.
I find this idea the work of genius. As well as offering premium-quality sound, the one medium would replace both analog cassette and CD. Not only would the record retailer be able to stock just a single inventory; also, the "Super CD" would offer customers something they haven't seen to any great extent since LPs' heyday: freedom of choice when it comes to quality.
Every HQAD purchaser could have the level of audio quality they're prepared to pay for. And should someone decide at a later date to upgrade their playback system, their investment in software wouldn't have been wasted. All they'd need to access the better sound is a new player. This combination of backward compatibility with CD and forward upgradeability maximizes the HQAD's commercial prospects—unless some mouth-breathing corporate knuckle-draggers decide against it. Or if the days of purchasing a physical item are over, and we'll all be downloading our digital music from the Internet instead.
Footnote 1: Since we purchased a large-screen TV, I'm astonished at how bad most broadcast-video quality is, compared, for example, with live sports coverage such as the recently finished 1995 World Series. (Welcome back to our home, Baseball!)