Fast Fore-Ward (Feb. 1997)

Every fall, the Audio Engineering Society holds its North American convention, which has two basic components. One is an exhibition of equipment from the various participating manufacturers of pro audio gear, ranging from elaborate mixing consoles for recording studios to loudspeaker systems for sound reinforcement.

Since pro audio is predominantly about sound production, rather than reproduction, the equipment tends to be correspondingly, and rather refreshingly, free-wheeling relative to the typical consumer goodies.

The exhibit hall had a charmingly split personality this year. On one side, the tech spiraled ever higher-advanced sound-editing software, DVD authoring systems, 24-bit A/D converters-while it took a distinctly retro turn on the other, with more tube gear in evidence than I can recall having seen before. My favorite item, however, was a little solid-state box from Aardvark, designed to give digital recordings the sound character of analog tape without the added noise (and inconvenience) of a trip through a real analog recorder. The AardScape, as it's called, emulates analog tape saturation; it's even adjustable, so you're not stuck with just one level or quality of "saturation."

The other half of the convention consists of workshops and presentations of research papers. Not surprisingly, a major topic this year was audio for DVD, both with and without video. Perhaps the best overview was a workshop on the first day, chaired by Tom Holman and featuring a half-dozen people heavily involved in DVD mastering. DVD is substantially more complex than any previous audio or audio/video format, requiring that multiple data streams-fixed-data-rate compressed audio (possibly in several languages), variable-data-rate compressed video, subtitles, and other ancillary data (including branching options)-be merged accurately and with precise synchronization.

The purpose of the workshop was to acquaint attendees with the intricacies of getting everything right the first time.

Two very encouraging pieces of information surfaced during this workshop. One is that Warner, Universal, and Sony-all represented on the panel-were deep into DVD mastering even at the time of the convention, in early November. The other is that the audio options for DVD-Movie, as it is known formally, are surprisingly rich. Dolby Digital (AC-3) will be the standard audio coding system for movies on DVD, but the specification also provides numerous linear PCM alternatives for discs containing less (or no) video-ranging from two 24-bit, 96-kHz channels to six 16-bit, 48-kHz channels.

The eventual DVD-Audio standard will provide additional capabilities-possibilities include provisions for more channels, lossless compression, longer words and higher sampling rates in the multichannel modes, and "Rosetta stone" data on such things as the loudness of the original performance and the acoustics of the recording site. But a great deal can be done within the DVD-Movie spec, which already represents a significant advance in audio capability over the decade-old CD standard.

The convention's anticlimax, at least for me, was Pacific Microsonics' paper on its HDCD process. It seems to boil down to this:

Initial conversion to digital is a very clean, 24-bit (19-bit actual resolution), 88.2-kHz process with subtractive dither. The resulting signal is then decimated to the CD sampling rate of 44.1 kHz via a digital filter whose coefficients are varied automatically according to signal content, in a manner that Pacific Microsonics says optimizes sound quality. These changes are tracked in a control code buried in the system's pseudo-random dither noise, which triggers complementary filter changes in HDCD decoders.

In reducing the signal to 16-bit, however, users of the encoding system also have the option to apply as much as 6 dB of peak limiting to signals that would otherwise clip and as much as 7.5 dB of upward compression to signals that range below -45 dBFS. As with the filter variations, these processes are tracked by codes embedded in the dither, enabling accurate restoration of dynamic range by HDCD decoders.

But what about undecoded playback? There's quite a bit of music that could be recorded with both processes turned full up and still never activate either. In that case, they won't make any difference. On the other hand, some music will emerge from undecoded playback with its dynamic range compromised. If that's okay with the artist and producer, it's their choice. I am disturbed, however, that Pacific Microsonics claims improved fidelity from HDCD encoding, including the dynamic-range manipulation, not only with proper decoding but also without decoding. Seems like a contradiction to me, and it leads me once again to ask, does whatever benefit the process might afford those who own HDCD decoders justify the loss to the much greater number who don't?

(adapted from Audio magazine, Feb. 1997)

= = = =