Project K622 Page 3
Faulkner: This will come as heresy to many, but I do not like 30ips as much as 15ips. For me, 30ips has more intrusive noise, more modulation noise, and a more eccentric bass replay response.
I used ¼" for several other reasons. First, because it is much more standard, given the disparate state of the analog recording industry, and left me more options for the mastering. So far I have not tracked down a ½" recorder, so I do not own one. Getting hold of new stock of the best analog blank tape these days is not trivial either, and there is more ¼" around than ½". Finally, ½" edits are more tricky than ¼" edits. I cannot escape from the necessity to edit, and ¼" gives me more flexibility.
Atkinson: And to my surprise, when I first walked into the control room, you were using Dolby-A noise reduction ahead of the Studer.
Faulkner: I used Dolby-A because classical music has a wide dynamic range, and setting levels is always guesswork, to some extent. Dolby-A gives me more dynamic range to work within—like using 24 bits in the studio for digital rather than 16 bits. Copying analog recordings always degrades the quality progressively, and Dolby A reduces the impact of any necessary copying.
Atkinson: Did it take some readjustment to balance the dynamic demands of the signal between analog tape overload and tape noise, hi-rez digital being more forgiving in this respect?
Faulkner: No. Hi-rez digital clips more offensively than analog. Both analog and hi-rez digital reproduce more dynamic range than many listeners are comfortable with, anyway. The Mozart Clarinet Concerto is not as much of a dynamic-range challenge as, say, Shostakovich 11—that would be more of a problem for analog.
Atkinson: And, as you said, you had to do your own analog tape splicing. Were there any splices that were routine in the digital domain that caused you some problems in the analog world? When assembling the performance from the session takes, I remember solving one problem that was going to involve you inserting a single turn in the clarinet at the end of a trill. This was trivially easy in digital space, but was probably going to have you breaking out in a cold sweat when you were taking the razor blade to the analog master.
Faulkner: Yes. Digital editing is a dream in comparison to analog editing. Once you are cutting bits of analog tape and piling on sticky tape, you cannot change your mind anywhere near as easily as you can with digital. With digits, you tend to do a rough first shot, then move things around till you are happy—after all that, you still have all the original material undamaged. All digital audio editors have got used to the luxuries of long cross-fades and dissimilar in/out cross-fades too, and you simply cannot achieve the same effects with a razor blade.
Atkinson: We did some comparisons in the control room. But given the choice between 24-bit/192kHz LPCM digital, DSD digital, and analog tape, which do you feel gets closest to the mike feed?
Faulkner: A well-made, first-generation analog recording using a good machine with good tape is a tough act to follow. The deviations from the live sound are there for sure, but tend to be both minor and, if anything, flattering—slight compression, some low-order distortion components, some warming of the bass. And a small amount of noise is no big deal in the grand scheme of things.
The question of which hi-rez digital medium is better, DSD or 4Fs PCM, is difficult. [4Fs = 4x44.1kHz or 4x48kHz—Ed.] Over the years I have grown to question DSD more, because of the dangerous quantities of EHF noise from the A/D noiseshaping. Apart from that (4Fs PCM is squeaky-clean in the EHF noise respect), I think there is little to choose sonically between DSD and high-speed PCM. The differences I do hear are down to fine details of filters and modulators. One thing is for sure: DSD is one pain in the neck for equipment designers.
DSD and high-speed PCM both sound very clean, but for me they lack the dynamics, warmth, and depth of both the direct sound and of the first-generation analog. The best analogy I can make is comparing a classic tube monoblock power amp—transformer-coupled, with very little loop negative feedback—with a squeaky-clean solid-state amp that uses huge amounts of feedback to obtain infinitesimally low distortion specifications. If you want an amplifier to create sinewave test tones, then the solid-state amp is supreme. But if you want to listen to Rachmaninoff, Sibelius, or Elvis, for me the classic old tube amp will communicate more dynamics and more music.
The question of overall feedback is an issue in the reproduction of musical dynamics, and modern A/D converters are heavily reliant on a feedback principle in their design, with a DAC in their [sigma-delta] loop. I think this is relevant.
Atkinson: DSD comes in for a lot of criticism in the academic community for its lack of theoretical elegance and for its demands on storage bandwidth compared with 24/96 PCM. Your feelings about those criticisms?