The Bell Tolls for CD Page 3

The PCM downmix doesn't just magically appear on the disc, of course. Sony calls this little trick SBM Direct—short for Super Bit Mapping Direct. Virtually any ADC or DAT recorder has a circuit that can downconvert 1 bit to 16 bits, but Sony wanted to maximize the advantages of DSD in a 16-bit wordlength. The answer was to again filter and noise-shape the DSD signal in a single stage—avoiding requantization errors and, according to Sony, minimizing aliasing and eliminating ripple. Thus, the SBM Direct processor was born: a one-stage FIR digital filter/noise shaper with 32,639 taps. You may already be familiar with what it can do and not even know it. Some commercial CD releases have used the SBM Direct processor and DSD downconversion and are on the shelves as we speak.

As with DVD-Audio, a little stuffing is necessary to get all of those tracks and extras onto the disc, even if the disc is dual-layered. For SACD, the compression happens courtesy of a Philips technology called Direct Stream Transfer (DST). Like MLP, DST is a lossless compression technique that offers a 2:1 data-reduction ratio with bit-for-bit accuracy to the original source. As you may remember, lossy compression systems like Dolby Digital, DTS, and all of the MPEG formats usually compress at a far higher ratio and cannot boast the same accuracy—although some are certainly better than others. These formats use perceptual coding and psycho-acoustic modeling to remove "unnecessary" elements of the soundtrack and further save space, a method that makes sense with movie soundtracks but (in my opinion) can be problematic for music. As with computer compression schemes, DST simply packs the information more efficiently, using adaptive prediction, data-framing, and entropy-encoding stages to accomplish its task. DST operates on a frame basis, enabling random access to the losslessly coded stream. Coding parameters are optimized once per frame. The entropy-encoding stages are adaptive and rely on probability: The shortest codes are assigned to the most frequent signals, while the longest codes are assigned to the least frequent signals. Of course, a closer analysis of the similarities and differences between DST and MLP will be in order for part three of this series.

The Last Bit
As with any new format, there are countless issues surrounding SACD that will have considerable effect on its final destiny. Improvements take time—and not just in the form of the medium itself, but throughout the entire chain. How will recording engineers deal with the new parameters they've been given? How about mixers, mastering engineers, producers, and the artists themselves? How will microphone technology improve to keep pace? Then there's the playback chain. How many amps, preamps, and speakers out there right now are prepared to deal with all of this extra bandwidth? Mark my words: You're going to see a lot more of the "super tweeter" concept in the future, as speaker manufacturers rethink tweeter design and explore the potential of increased frequency response.

As always, there's also the ubiquitous issue of copyrighting. SACD addresses this issue with both visible and invisible watermarking systems, which unfortunately may be audible (according to some experts). One of the protection technologies being used is called Pit Signal Processing (PSP)—which can put a faint image, in the form of text or graphics, on the signal side of the disc, making it hard for pirates to duplicate the disc without visibly corrupting the watermark. Naturally, there are fears about how these signals will affect the audio signal and whether or not they're as bulletproof as their creators would lead us to believe. This latter fear will certainly not inspire studios to release their full catalogs in high-resolution. Of course, the potential for digital output of SACD from the player will hinge greatly on these copyright issues.

As for SACD's sound—the only part that really matters at the end of the day—I'll once again leave the main exploration of that to the player reviews in this issue. I could go on and on about the deep, incredibly well-defined soundstage, the silky top end, the dead-on accuracy, the timbre of the instruments and voices, and so on. However, all I'll say here is this: I agree that SACD is probably the most analog-sounding digital format I've ever heard—and that's saying a mouthful.

It's ironic that, on the surface, the compass of progress points squarely into the past: The best we can hope for from our digital systems is to emulate the best of our analog systems. When you remember, though, that analog is a physical/electrical representation of the original experience and that digital is a processed representation of analog, it starts to make more sense. No one ever said (or should have ever said) that analog didn't sound good; it's just that it has many aspects that, for lack of a better phrase, are a pain in the ass. If SACD and the other high-resolution formats can successfully combine the ease and flexibility of digital with the sound of analog—while at the same time capturing the hearts of the masses and not getting bogged down by their own peripherals—then the future of audio really is upon us. CD, MP3, and the rest of the usual suspects may be alive and kicking for now, but their graves have been dug. What actually gets put in those graves is up to you.