Festival! The Best of the 1995 Santa Fe Chamber Music Festival Recorders & Post-Production

Recorders & Post-Production: John Atkinson

We were very fortunate to be able to use a Nagra D digital recorder for the Festival project. This Swiss open-reel recorder packs a huge amount of superbly engineered, computer-controlled functionality within its tiny frame. Widely used for film-sound location recording, it is capable of storing either two or four channels of up to 24-bit digital data on 5.75" reels of Ampex 467 tape. In two-channel mode, each reel lasts two hours; four data channels give just over an hour's worth of recording time at a 44.1kHz sampling frequency (48kHz is also available).

As as well as being able to interface with just about every video machine made, the Nagra has four channels of both digital and analog input/output ports. I did some tests comparing the Nagra's internal 20-bit A/D converters with Stereophile's 20-bit Manley ADC---the one we'd used for Concert. While there was a very slight tonal difference, there was no overall quality difference that I could hear. So it made sense to go with the Nagra, particularly as the tape recorder includes four very-low-noise/very-high-headroom microphone preamps that could also provide the 48V phantom power required by the B&Ks.

Although the Nagra has conventional-looking peak-level meters---quaintly called "modulometers"---I also monitored recording levels with a Dorrough AES/EBU meter. This useful instrument accepts a digital data input and has a 1dB-resolution scale from -1dBFS down to -30dBFS, with -2LSBs, -1LSBs, and 0dBFS indicated by the top three LEDs turning red. I set the levels as high as possible to maximize signal resolution, allowing just 2dB to accommodate the possibility of a musician playing that bit louder live than at rehearsal---during each performance, I hardly dared to breathe in case we clipped an A/D converter. We didn't---but not without a good deal of nail-biting.

For on-site monitoring, Wes Phillips and I used headphones exclusively (though I critically listened to each night's work the next day on my usual reference system of B&W Silver Signature speakers driven by Mark Levinson electronics). I mainly used Stax SR-Lambda Pros fed by a Meridian 263 D/A processor, supplemented by a pair of Sennheiser HD 580s, while Wes used the Limited Edition Anniversary edition of the Sennheiser HD 580 Jubilees.

For each piece of music I recorded three performances, including the sound check. In this manner, though I intended to use as much of a complete performance as possible for each work, we would be covered if there were any wrong notes or objectionable audience noises. For Ms. Kohjiba's The Transmigration of the Soul, which starts with an unaccompanied soprano (Kendra Colton) at the very back of the hall behind the audience and has several passages in which the contrast between quiet instrumental sounds and silence is fundamental to the musical meaning, we rerecorded about five minutes of music after the audience had left the hall following the second performance. This was unavoidable, given the propensity for even the quietest audience to be noticeably noisy. I don't believe this compromises the integrity of the "live" recording philosophy.

For the editing, all four 20-bit tracks were uploaded to a Sonic Solutions hard-disk editing system running on a Macintosh computer. As well as control of individual track level and polarity, the Sonic Solutions software allows the time relationship between tracks to be adjusted. This facility was essential for this project, as the outputs of the omni microphone pair and the cardioid microphone pair were not in time-sync. As explained above, I'd chosen the exact positioning for each for reasons of tonal balance and soundstage optimization.

The first task, therefore, was to slide one pair of tracks along in time to achieve synchronization with the other. To do this optimally, I had recorded Wes Phillips standing just behind the conductor's podium and banging a pair of slapsticks. By looking at the transient waveform on each of the four tracks, I could adjust the relative timing of each so that the overall mix would be phase-coherent. As a result, any disparity between the time stereo image from the omnis and the amplitude/time stereo image from the cardioids would be minimized.

Once Heiichiro Ohyama and I had assembled a master edit list for the CD, I had to decide how to reduce that 20-bit data to the 16 bits mandated by the Compact Disc Standard. Simply dumping the output of the Sonic Solutions hard disks to DAT or CD-R, thereby truncating each digital word from 20 to 16 bits, both reduced the sense of recorded space and added a feeling of "digititis." Following my positive experience of the Meridian 618 Mastering Converter used on Concert, I used Meridian's new 518 machine. This uses a Motorola DSP56002 digital signal processing chip to manipulate digital data in real time. The input and output data word lengths can be independently selected; pre-emphasis or gain can be added if the operator so wishes; and, most important, the 518 redithers the data with a choice of noise-shaping curves. By shifting quantizing noise up to the inaudible 20kHz region as it reduces the output word length, it preserves as much as possible of the original's resolution in the midrange. Using the Meridian's noise-shaping algorithm when we downloaded the 20-bit master to a 16-bit CD-R gave a noticeable improvement in air, space, and overall palpability.

I believe Festival captures the vivid sense of excitement and tension of great live performances. If playing this recording makes you reach into your pocket to check if you still have the ticket stub, we'll count it a success. Festival costs $15.95 plus S&H. See the secure "Recordings page" to order it.

Warning: This CD preserves the wide dynamic range of live orchestral sound. The Copland work starts very quietly; be sure not to set your volume control too high or the climaxes in the Kohjiba and Milhaud works might knock you backwards in your chair!