Deep River: the Cantus Spirituals Project Page 2

The outputs of the DPA 4003 omnis were brought up to line level with their power supply-preamp then fed to two channels of an eight-channel Metric Halo MIO 2882—an immensely versatile, FireWire-controlled preamp and A/D converter. This was hooked up to an Apple Titanium PowerBook G4, and my original plan was to record the two omni channels straight to the laptop's hard drive using Mac OS X and the Bias Peak 3.0 program. However, I could not get a reliable lock with the program clocked from the MIO 2882, which in turn was slaved to the master dCS converter. While the MIO had no trouble locking to the incoming word clock, Peak would change the MIO's sample rate from 88.2kHz to 44.1kHz when I pressed Stop even though it, Core Audio, and the MIO were all set to 88.2kHz. (The 4.0 update of Peak, which I subsequently purchased, does not have a problem in this regard.)

My workaround at the session was to use the PowerBook for control and signal monitoring, using the MIO's AES/EBU digital output to feed data to an RME Digi96/8 Pro soundcard mounted in a PC. This was a Pentium 4 machine sourced from hot-rod computer company UniQpc running Cool Edit Pro (now called Adobe Audition); this stored the data on a LaCie 120GB FireWire hard drive. The PC used the tubed Pentium IV motherboard from A-Open, which was used some of the time for monitoring. The rest of the time I used a Musical Fidelity X-24k DAC driving HeadRoom BlockHead, Musical Fidelity XCans V2, and Stax SRM-T1 headphone amplifiers. My headphones were Sennheiser HD580s and '600s, Stax Lambda Pro Signatures, and Sony MDR-7506s.

If my combination of heterogenous tape recorders, converters, and preamps seems an overly convoluted setup—I could have used the Millennias also to feed the MIO 2822 and used a multichannel program like Nuendo on the PowerBook—if there is one thing I have learned over years of recording—okay, it's two things—1) always use whatever worked the last time, and 2) never rely totally on something that you have not used before when the session clock is ticking.

In theory, the notebook-sized MIO, hooked up to a laptop, is all anyone would need for eight channels of high-quality 24/88.2 recording (footnote 1), but as I was still going to run tape as a backup, I was not going to reduce complexity. And an hitherto unappreciated virtue of tape is that, unlike a file on a hard drive, it can't be easily erased! My setup may be convoluted, but it has served me well in the past. The ADCs sync up as expected, the Millennias are ultra-quiet, gremlins know to keep their distance, and the sessions go smoothly. (Smoothly, that is, apart from traffic, planes, police helicopters, distant trains, squeaky piano pedals, air-conditioning that the custodian insists can't be turned off, birds, and once even a bat that kept circling the microphones.)

For the Sioux Falls sessions, the MIO 2882 and one of the dCS converters were slaved to the other dCS converter so that all three pairs of microphone channels were converted to digital with the same word clock. This way, when the data were loaded into a digital audio workstation, they would be synchronous and could be edited and mixed without phasing problems. As with all of my recordings, I could slide the microphone pairs backward and forward in time in post-production, my goal being to present a stable, well-defined soundstage while preserving the characteristic benefits of each pair. To facilitate this process, each take began with the singer standing at the precise center of the stage swinging a slapstick to give a timing reference for all six channels.

The necessary 88.2kHz-to-44.1kHz downsampling for the CD master was performed with a dCS 972 D/D converter, while the 24-bit master mixes were dithered to the 16 bits required by the CD medium with the excellent-sounding POW-R algorithm, this running on the Z-Systems rdp-1 digital equalizer-preamplifier that I purchased following Kal Rubinson's rave review in July 1999.

No compression was used in the preparation of the master. However, I did apply judicious equalization, both with Cool Edit Pro and with the rdp-1, which is about as sonically transparent as you could wish. Reflecting how technological developments democratize the making of recordings, I made the CD master on my PC, using a Plextor Premium CD burner and Sony's CD Architect program. The beauty of the Plextor drive is that it comes with a suite of software that allows you to analyze the integrity of the burned disc, monitor time-base error (jitter), log the incidence of corrected and uncorrected errors, and even examine the physical state of the pits. You can therefore burn and reburn, experimenting with different media and write speeds, until you get a disc you can send to the pressing plant that has no uncorrected errors and minimal corrected data errors, thus the maximum possibility of sounding as good as the original.

I'll give the final word on the project to Erick Lichte: "We hope that this recording will serve as the definitive recording of these male-voice spiritual arrangements. It is our hope that the person who buys Deep River because it seems like pleasant and familiar music should, after hearing it a couple of times, understand the narrative of the African-American experience in this country in a subconscious way or even, perhaps, consciously."

Deep River costs $16.95 plus S&H and can be ordered from our secure "Recordings" page.

Footnote 1: The MIO 2882's mike preamps are not quiet enough for distant classical miking. (For that purpose, Metric Halo makes a two-channel FireWire model, the ULN-2.)