DASH-ING THROUGH THE SHOW
It was autumn in New York again, and the 74th Convention of the
Audio Engineering Society held forth October 8th through 12th. However, instead
of being in the familiar, warm ambience of the Waldorf-Astoria, it was convened
in the starkly commercial precincts of the New York Hilton.
Not many people were ecstatically happy about this change in venue, including yours truly; one complaint will serve to demonstrate our frustration. In the world of the New York Hilton, with its bottom-line-oriented efficiency, on the entire lobby floor and in the huge anterooms to the exhibit spaces on the second and third floors, there was not a single, solitary chair on which one could sit without encountering a cocktail waitress. Since informal "bull sessions" among AES members have always been a very important social function and a means to discuss and disseminate technical information, the lack of chairs torpedoed this activity. The only recourse was to the strategically placed bar/lounges, where if you didn't mind paying $1.75 for a coke or $2.00 for a cup of coffee, you could frolic with your peers.
Still, life does go on, and this was a very significant convention. Among other things, there was so much activity in the burgeoning world of digital audio that the convention ran to five days for the first time in its history.
Saturday, October 8th had been designated "D" (digital) Day, and was wholly devoted to digital technology and workshops. All Saturday afternoon there was a hands-on digital workshop co-chaired by my recording associate Frank Dickinson and Jonathan Howard. Frank demonstrated the use of the professional 2-track JVC digital recorder and the 32-track 3M digital recorder for digital recording and editing techniques. Jonathan demonstrated how the Sony PCM-F1, with the aid of several black boxes, could be made compatible with the professional Sony 1610 digital recorder and also synchronize with SMPTE time code and video interfaces. There was also an interesting day-long digital tutorial on Saturday which was very well received and gave a grounding in basic digital audio for those not yet familiar with this technology.
In my opinion, there were a number of papers presented which could be of landmark significance. Surely qualifying in this respect was "Signal Enhancement via Digital Signal Processing" by Dr. Roger Lagadec and D. Pelloni of Studer. Dr. Lagadec is a gifted young scientist, quite possibly the brightest star in the digital firmament.
In fact, he deservedly was made a Fellow of the AES at the Convention Awards Banquet. The signal enhancement Dr. Lagadec has come up with is nothing less than the removal of noise from music program sources, plus the distinct possibility that, in time, such things as intermodulation distortion may be reduced. In Dr. Lagadec's own words: "Signal enhancement describes attempts at approximating original musical signals from non-ideal recordings.
In a novel processing method, a digitized audio signal is split into phase linear, complementary narrow-band signals for individual processing prior to being re-merged to form the enhanced output signal.
"It is intuitively appealing to use digital audio techniques not only in the digital recording and processing of today's signals, but also in the archiving and possible restoration of yesterday's valuable musical heritage. The task is almost a contradiction in terms: Possessing a recording of an unknown musical signal degraded by unknown noise components, one wants to preserve the music and suppress the noise, possibly in real-time, and with only a modicum of manual control.
However, whatever in our brains differentiates between 'music' and 'noise' does not yet accept straightforward modeling. Some examples, however, indicate that musical signal enhancement should be achievable.
"Two different applications will be considered: One is to take decent recordings and improve the SNR by a perceptible amount, so as to make them better suited for new media such as the Compact Disc. The other is to take recordings of evident cultural value but objectionable technical quality and attempt to restore what occurred in front of the microphone. In the first application, noise components are small, and already far down the level scale, although still perceptible. In the other, they are large enough to interfere with the musical signals." To accomplish this signal enhancement and noise reduction, Dr. Lagadec has designed a very special 512 band digital filter, far too complex to describe here. This multi-band filter has been used in a broad range of processing and enhancement experiments. Dr. Lagadec describes some of his results: "A large number of takes have been processed and stored digitally, ranging from high-quality analog masters where only very low noise components exist, to precious, but noisy early recordings, and to artificial signals, as used in assessing digital audio. Some results are humbling: Telling a computer to remove very much noise but to preserve a pianist's humming and breathing, while possibly removing some distortion effects--the whole of it on signals read from an LP-indicates enthusiasm, but does not lead to fully convincing results. Other results are extremely encouraging. Most results indicate that the approach is powerful, the parameters many, and that only professionals with musical expertise will know how to benefit from the system, and how to adjust the parameters properly." Dr. Lagadec goes on to say that "An approach to signal enhancement based on efficient filter bank design has been presented. It is based on consequent refinements of the already powerful polyphase method, aimed at making it useful in digital audio applications. The result is a very high quality real-time, time-domain filter bank with linear phase, which can be implemented in hardware." Brought down to practical aspects, Dr. Lagadec's multi-band enhancement/noise-reduction filter opens up fantastic potentials in many areas. Apparently, under optimum conditions the device could be used to remove the residual hiss from thousands of pre-Dolby tape masters (and, of course, from Dolby-encoded tapes as well). The late Bob Fine's famous Mercury Olympian Series recordings immediately come to mind. So do my own Everest three-channel half-inch tapes and 35-mm magnetic film masters. Think of the music treasures in the vaults of Decca, EMI, Philips, DGG, RCA, CBS, etc.! All this enhancement presumably without degradation of the music signals! Dr. Lagadec was kind enough to demonstrate his filter system in his hotel suite. Even though it was a fairly primitive example of this new technology, the results were nonetheless impressive. The reference a few paragraphs earlier to the "pianist humming" was an old 1955 Glenn Gould LP. The improvement between the unprocessed and processed sound was striking. Hiss certainly was reduced, but more than that there was a reduction in distortion. As Dr. Lagadec pointed out, there is still a long way to go, but I'm betting he will nurture the device and develop it to its full potential.
During the convention, a joint press conference was held by Sony, MCI, Matsushita and Studer so they could introduce us to something called "DASH." Is this a new soap powder? No, friends. DASH is the acronym for Digital Audio Stationary Head, a digital recording format jointly adopted by this group of companies.
Essentially, this is an attempt to set a standard for digital recording which will be internationally adopted by the audio industry. I believe that the idea and the goals are laudable. Full compatibility between the digital recorders of the various companies would mean easy international tape interchange.
Digital copying could be direct, with no necessity of leaving the digital domain.
There are three versions of the DASH format, depending on fast, medium and slow tape speeds. However, encoders and decoders of all versions are identical. The number of tracks required to record one digital audio channel also varies. All three tape speeds operate at the 48-kHz sampling rate, which is also recommended as an international standard. The version of DASH for fast speed operates at 30 ips, requires one track per channel, and can accommodate 48 digital audio channels on half-inch tape. Medium-speed (15 ips) DASH requires two tracks per channel and can accommodate eight digital audio channels on quarter-inch tape, and 24 channels on half-inch tape. Finally, the slow-speed version, at 7 1/2 ips, requires four tracks per digital audio channel, and accommodates two or four channels on quarter-inch tape.
Few would argue that the DASH format is not a good idea. In the best of all possible worlds, DASH would be adopted and we would all start digital recording, say, for example, on January 1, 1984. Unfortunately, there are thousands of reels and cassettes of digital recordings made in the last few years on Sony, Soundstream, JVC, 3M and Mitsubishi recorders. None of these machines are compatible with each other, nor can any of them make direct digital copies to DASH recorders, but must convert D-to-A and then back to A-to-D. This means, in essence, the present digital recorders must be maintained to service existing digital libraries. Is there a solution? Apparently none yet, and as long as this situation continues, it will certainly delay the adoption of DASH.
-----------
(adapted from Audio magazine, Jan. 1984; Bert Whyte )
= = = =