COMPRESSION IMPRESSION
To apply the term "high fidelity" to the reproduction of recorded music implies that the sound is a completely faithful image of the live musical event. Since the beginning of the hi-fi era, audiophiles have pursued the elusive goal of achieving true high fidelity sound. They have employed the most extraordinary measures to extend response at both extremes of the audio spectrum, reduce all forms of distortion to insignificance, achieve flat frequency response, and, in every way, preserve the integrity of music signals.
Over the years, advances in recording technology were matched by improvements in playback equipment. The vinyl LP record was the principal music storage medium, while the more affluent favored music on prerecorded open-reel tape. As technology progressed, there was a transition from tube electronics to solid-state designs.
The advent of Dolby noise reduction was a most significant development, both for professional recording and for its contribution towards making cassette tape a respectable source of re corded music. With the introduction of the Compact Disc, digital audio be came the cutting edge of the art.
Once these technological advances demonstrated their value in bringing us closer to musical reality, they earned the support of audiophiles and were integrated into the audio scene. For more than 40 years, steady incremental improvements towards true high fidelity music reproduction have brought the art to its present level of sophistication. Having achieved this, the audiophile fraternity zealously guards against any idea or technology that would subvert or degrade these hard-won standards. Among many of the "top gun" audiophiles, signal processing and manipulation of any kind have been anathema. These people are disdainful of any kind of equalization, reverberant processing, and even noise reduction. Quite often, they are avowed digiphobes. (Of course, we all occasionally agree on the deleterious effects of things like Copy-Code!) All the foregoing is significant in light of several new developments in digital technology which could seriously com promise some cherished audio tenets as well as cause divisiveness and controversy among audiophiles. By now, most readers are aware of the Philips Digital Compact Cassette (DCC) and Sony Mini Disc (MD) formats. Both use data compression techniques, which are not used in the CD format. To sonic purists, psychoacoustics and masking are dirty words. Since the DCC and MD data compression techniques involve these phenomena, many audiophiles simply dismiss DCC and MD as unworthy of consideration.
With the old adage of "Ain't science wonderful?" in mind, it appears that progress has been so rapid in digital audio engineering, the much-maligned concept of data compression has gained new respectability. This is especially true in regard to DCC. Unlike rotary-head R-DAT, DCC uses thin-film head technology to record and play back digital data. A bit-rate reduction system is used, employing what Philips calls Precision Adaptive Sub-band Coding (PASC). It is a very efficient coding system, needing less than a quarter of the bit rate of CD.
Barry Fox, one of Britain's most erudite and perceptive audio journalists, provided a particularly lucid and de tailed description of how PASC works for Hi-Fi News & Record Review (March 1992):
The audio frequency range is split, by digital filtering, into 32 sub-bands of equal width. The PASC signal processor models the aural characteristics of the spiral cochlea inside the ear, taking the threshold of hearing (of the most sensitive human ear) as its basic reference.
As the sound is registered, the signal processor continuously adapts to the dynamic variations of the threshold. Efficiency is increased since only sounds above this dynamic threshold, and thus audible, are accepted for coding. Every sub-band is allocated the bit capacity it needs. The bits not required by particular sub-bands are re-allocated, adaptively, to other sub-bands to achieve the highest possible coding accuracy over the entire audio frequency range.
Hence, PASC coding takes the behavior of the ear as the reference model, and uses the chip technology of the 1990s to emulate that model.
The coded information for all the sub-bands is multiplexed into an 8-channel data stream, into which error detection and correction codes are inserted. The distribution of the block codes over the 8 data channels achieves an effect similar to interleaving. A ninth channel, carrying the control and display auxiliary channel, is added. 8-10 modulation, to optimize the data format for recording on tape, then completes the signal processing.
Naturally, with such a complex and sophisticated concept as PASO, the first thing most people want to know is if it in any way degrades any parameter of audio quality: Are any artifacts audible? In the initial stages of DCC development, the effect of PASO was clearly audible seven or eight times out of 10. Philips is planning a world launch of DCC recorders and at least 500 titles of prerecorded DCCs in September.
With the latest embodiment of PASO, Philips wanted assurance that this sys tem would not produce detectable audio glitches and that it was sonically transparent. Thus, they enlisted the golden ears of the engineers at Decca Records (conveniently part of Philips' PolyGram group). To me, this made a lot of sense because of Decca's musical and engineering expertise, built up over many years.
In 1976, I was the guest of Decca's great pioneering engineer, the late Arthur Haddy, and engineer Tony Griffiths, now general manager of the Dec ca Recording Centre in London. At that early date, they demonstrated Decca's proprietary digital recorder to me.
From the beginning, they used a sampling rate of 48 kHz (now the professional standard), and the recorder was capable of 18-bit performance (al though Decca was using it for 16-bit coding). In 1991, they switched to 18 bit recording and, with the company's extensive recording schedule, now have a large supply of 18-bit masters on hand. I was Decca's guest on several other occasions, most recently in 1986. At that time, I had the pleasure of meeting Jim Lock, the chief recording engineer, and John Dunkerley, the Decca engineer who has made so many memorable recordings.
Recently, I had a phone conversation with Dunkerley in which we discussed a series of tests Griffiths had run on a Philips DCC recorder, paying particular attention to any glitches from the PASO system. In Griffiths' test, a Decca 18-bit master tape was output to the DCC recorder, which was set up for 18-bit encode/decode. An A/B switch was used to compare the master with the DCC decoded output to determine if PASO was in any way detectable. The Decca engineers (including Jonathan Stokes, who had worked with earlier PASO units) could not reliably state whether they were listening to the master or to the decoded DCC! As Griffiths said, "We may think we hear something, some of the time, but on over a hundred tests, there was no statistical evidence that anything was audible." Now, Tony Griffiths is a brilliant but basically no-nonsense, conservative engineer. He is so impressed with the results of the DCC tests and performance of PASO that he thinks a DCC recording from a Decca master could be used as the dubbing master for prerecorded DCCs! He says that DCC now has a dynamic range greater than CD's! I asked Dunkerley what he felt about the NB tests. On that kind of short-term basis, he replied, the results were convincing and impressive, but he said he wanted to hear some DCC recordings in normal, extended listening conditions.
Philips plans to introduce entry-level DCC recorders with 16-bit performance, with Marantz introducing a more upscale 18-bit DCC recorder.
Griffiths feels that if a 20-bit D/A chip is made available, Decca could provide 20-bit masters for a complete 20-bit DCC recorder! So, audio purists, what now? Has the rapid advance of digital technology finally done the unthinkable? Has it made data compression undetectable? Just remember, there is digital data compression for video too--all the HDTV systems under FCC consideration use it-so perhaps it is a technology whose time has come.
(adapted from Audio magazine, Jun. 1992; Bert Whyte)
= = = =