MORE SIGNIFICANT BITS
No matter what medium is employed, there can be no high-fidelity reproduction of music unless there is a corresponding high-fidelity recording of the music. This is axiomatic and immutable. The most elaborate and sophisticated audio component system cannot make a poor recording sound good; as we all know, the higher the quality of the playback system, the more we will hear the sonic warts of a poor recording.
After World War II, a certain amount of progress was made toward the improvement of sound quality in 78-rpm recordings. The culmination of all this was that in 1948, Decca introduced their famous ffrr (full frequency range recording), surely the zenith of 78-rpm sound quality. The parallel development of magnetic tape recording and the 33 1/3-rpm microgroove long-playing vinyl record ushered in the high-fidelity era in 1949.
From the beginning, open-reel magnetic tape recording provided a medium that, at 15 ips, had a wide frequency range (30 Hz to 15 kHz) and a dynamic range and S/N ratio of about 55 dB. It also afforded a recording time of 30 minutes at 15 ips, and the luxury of tape editing.
Of course, it soon was realized that improvements in recording quality could be well publicized and result in increased record sales. In the early 1950s, considerable advances were made in magnetic oxide formulation magnetic head structure, and recording electronics. Condenser micro phones, as exemplified by the famous Telefunken U-47, were increasingly used. Virtually all the record companies utilized some or all of this technology, but recording advances were most assiduously pursued by the "sound labels" of the day: London Decca, Westminster, Vanguard, and of course, Bob Fine's highly acclaimed Mercury Olympian series.
This was still the era of monophonic recording on quarter-inch tape. As the quality of the tape masters improved so did the need for better disc-cutting heads and electronics. I well remember Bob Fine first using a Grampian cutter and then adopting the very high quality Miller cutter, which required a hulking McIntosh 200-watt tube amplifier to drive it optimally.
When it became apparent that stereo recording would soon supplant mono recording, Bob Fine wanted to record three channels of sound on 1-inch tape. The tape companies believed they would have trouble maintaining uniform slitting at this width and suggested the use of three channels on half-inch tape, which also made fo a less expensive tape head structure. Subsequently, Ampex marketed the 300-3 recorder in the half-inch format, but obviously, the S/N ratio suffered because of the reduction of track width in using half-inch instead of 1-inch tape.
With the advent of the stereo disc in 1958, three-channel stereo mastering on half-inch tape grew enormously, followed later by four-channel recording on half-inch tape and multi-channel recording on 1- and 2-inch tape. During this period, recording engineers were trying to eke out every last iota of sound quality with respect to extended frequency response and dynamic range, lower distortion, and better S/N ratio. Much of this was accomplished with better and quieter tubes, more refined mixing consoles, and more sensitive microphones. The stereo discs benefited from more linear Neumann and Ortofon cutter heads. The sound labels tried to stay with the simpler, "purist" style of stereo microphone techniques, while the major labels started to apply multi-microphone stereo recording to classical music. Thus, anything and everything was used to improve sound quality and gain a competitive sales advantage. The sound oriented labels had to put their emphasis on high-quality sonics because their artists were not of the illustrious stature of those on the major labels.
On my Everest recordings, we initially used three-channel half-inch Ampex recorders with considerably modified and improved recording electronics. This was easy to do since we could use the very sophisticated facilities of our parent company, Belock Instrument Corp. To push Everest to the technological forefront, we took the expensive plunge into three- and six channel mastering on 35-mm sprocketed magnetic film. Now, with three channels-each the equivalent of full width, quarter-inch, single tracks-we were routinely achieving S/N ratios of around 70 dB. With the film running at 90 feet per minute, this was equivalent to 18 ips and afforded extended high frequency response.
An unfortunate aspect of analog magnetic tape recording is that a copy of the master tape will be degraded in various ways, especially in respect to S/N ratio. The problem is alleviated somewhat with Dolby A NR, but in general, copies of analog master tapes just are not as good as the original.
What really prompted this month's column is that now, in 1989, 31 years after Everest started using 35-mm magnetic film for mastering, a seemingly impossible thing has happened. Imagine my surprise to find that Philips recently has licensed some of my Everest recordings: Aaron Copland conducting his "Billy the Kid" suite and Third Symphony, Stokowski conducting the Shostakovich Fifth Symphony, Ferde Grofé conducting his "Grand Canyon Suite," and Carlos Chávez conducting some of his works. Philips had the Everest 35-mm magnetic-film masters processed at Sonic Solutions, using their NoNoise digital noise-reduction system. The recordings were subsequently issued as Philips "Legendary Classics" on CD. Incredibly, the CDs are quieter than the original 35-mm masters! Evidently, the 70-dB S/N ratio of the 35-mm masters made these tapes particularly suitable for the NoNoise digital processing. I can tell you, it was one helluva thrill to hear my recordings not only resurrected but sounding better than ever! Nothing has been changed or degraded in any way. In fact, the absence of noise provides better detail and resolution and affords a deeper insight into the music.
To put some icing on the cake, it should be noted that Bob Fine made many Mercury recordings on 35-mm magnetic film and that Philips owns these masters. It would seem reasonable to expect that some titles will be issued as Legendary Classics CDs.
It was quite a few years after the Everest 35-mm masters before conventional magnetic tape recording--with the addition of Dolby A NR--could equal, and then surpass, the S/N ratio of the Everest tapes.
I have pointed out before that digital recording is the great leveler. No matter what brand of digital recorder is used, the performance is very similar.
The same holds true for the CD. With few exceptions, if a CD plant follows the Sony/Philips manufacturing protocol, the CDs will be like peas in a pod.
In audio engineering, progress rarely slows for very long, and the winds of change are blowing again. By general agreement, one aspect of digital re cording that needs a technological update is A/D conversion. With eight times oversampling common on the D/A converters in many consumer CD players and 64-times oversampling D/A converters available on specialized processors, it is surprising to learn that two-times oversampling is generally supplied for A/D conversion on professional digital recorders.
Recently, several new IC chips for ND conversion have become available. Among these is one from dbx/ CTI: An 18-bit A/D converter operating at 6 MHz. Theoretically, this is equivalent to 128-times oversampling, a fact Chesky Records, who uses the dbx/ CTI chip, proudly proclaims on the covers of their new jazz CDs. This IC chip is also being used for some Telarc recordings. In a conversation with Tony Griffiths, head of Decca recording in London, I was told that there is a new 20-bit version of the dbx/CTI chip, and that with Decca's proprietary digital recorder, he is getting 19 1/2-bit performance! In addition, it appears that Sony is introducing 20-bit A/D conversion chips. According to audio super-sleuth Barry Fox, a Sony 3402 DASH format digital recorder, using quarter-inch tape at 15 ips, has a modified head track pattern necessary for 20-bit recording. On this experimental unit, the ND conversion electronics are mounted in an external black box but will be incorporated in the chassis of the production recorder, of course. Since the CD is a 16-bit system, why use 20 bits? It is somewhat analogous to headroom, with the 20-bit recording ensuring total resolution of the PCM system's 16 bits. As most readers know by now, Sony bought CBS Records; it appears they are founding a new label, "Sony Classical." These recordings will be designated "20-bit digital." Apparently, the EMI mobile recording truck had the experimental 20-bit Sony recorder aboard, so the unit was used to record Dietrich Fischer-Dieskau on some Mahler songs with the Berlin Philharmonic, and to record the Mozart Requiem with the London Philharmonic in Walthamstow Town Hall.
This appears to be quite an ambitious undertaking for Sony. The 20-bit ND converter will probably be shown at the 87th AES Convention in New York City. Although this 20-bit recording technology would give the new Sony label a good deal of technical one-upmanship, I rather doubt Sony will use it exclusively for very long.
In any case, even if new developments in digital recording give some companies a technical advantage, there will be a return to digital equality in fairly short order. In the long run, the sonic quality of digital recordings still depends on the skills of the recording engineer, on the microphones he uses, and on the manner in which these mikes are employed after careful consideration of the acoustics.
(adapted from Audio magazine, Dec. 1989; Bert Whyte)
= = = =