Features

Sidebar 3: Just a Little Echo, Echo, Echo...
Back in my recording-musician days in the '70s, the studio at which I was the house bass player had an AKG reverberation chamber that found its way onto pretty much everything we laid down on tape. Basically a large metal plate with an electromagnetic exciter at one end and a pickup at the other, this produced reverberation with a not unpleasant sonic signature. In conjunction with discrete tape-delay echoes courtesy of a high-speed ReVox A-77, the effect could be tweaked to give quite a lifelike simulation of a concert hall—or at least one that passed muster for the rock recordings we were producing. And when I worked on an album at EMI's famed Abbey Road studios, I was intrigued to learn that they used real reverberation chambers, with a loudspeaker at one end and a microphone at the other. The speaker was on rails, allowing it to be moved closer to the mike to reduce the apparent size of the room.

These days everything is done in the digital domain, and relatively low-cost digital reverberators have revolutionized studios. But as I got more involved in classical recording, it seemed obvious that the role of the engineer was to capture not just the sound of each instrument as accurately as possible, but also the sound of the room. The result would be a recording that, when played back on a system with inherently good imaging performance, would throw a simulacrum of the original event between and behind the speaker positions—a soundstage! My purist attitude was this: If you want to make a great-sounding classical recording, first find a good-sounding room.

And that we managed to do with Stereophile's first 10 recordings. But, as Wes Phillips describes, we hit an impasse with the Encore project: the stage setup for the 1997 Santa Fe Chamber Music Festival had been changed, optimizing the sound for the audience at the expense of the microphones. Moving the microphones back was not an option; they then would have picked up too much slap-echo from facing frescos on the auditorium's sidewalls. And in fact, for the Mendelssohn Sextet, we actually had to place the microphones even closer than for the Brahms, due to the demands of video engineers who were taping the world premiere of the other work on that night's program: Feliu Gasull's Contra-xions for flamenco guitar, instrumentalists, and two dancers, one of the Festival's two commissions for its 25th-anniversary season.

As a result, the recorded sound we ended up with was immediate and vivid, but there just wasn't enough of the room. There was also a puzzle to be pondered: the 24-bit tapes sounded acceptable, and I got used to the presentation when I was assembling the master. It was only when I redithered and noise-shaped the edited master file's word length to 16 bits that the dryness of the sound became subjectively bothersome.

Should I add some artificial reverberation? That was the question.

Attending Audio Engineering Society conventions over the years, I have always tried to catch presentations given by Lexicon's founder, David Griesinger. A fascinating speaker and a keen recordist, Griesinger has probably done more analysis of reflections in real rooms and more closely investigated the perception of reverberation than any other engineer. His Lexicon Model 480L digital reverberator, for example, is a standard piece of kit for classical engineers with deep pockets. So when I attended the 103rd AES convention in Manhattan last September, I stopped by the Lexicon stand and chatted with the Massachusetts company's then VP of North American Sales, Joel Silverman. "Check out our new $3000 PCM 90 Digital Reverberator," advised Mr. Silverman. So I did.

Lexicon's DSP-based PCM 90 offers 250 preset reverberation programs, many based on real halls, many synthesizing hall acoustics that could not happen in reality. Many of the parameters of each of these presets are adjustable—stage width, for example, or the overall decay time, or the manner in which high frequencies are absorbed. As well as conventional analog inputs and outputs, the '90's S/PDIF ports allow it to work on digital data. While its internal data processing operates at 20/24 bits, the overall resolution is 18 bits. As long as I operated the unit 12dB hotter than I would eventually need, I could achieve effective 20-bit resolution by backing off the gain on the data by the same 12dB.

One key to making the Encore sweetening work, I felt, was to find a PCM 90 reverberation setting that most closely resembled the sound of the real hall. During the performances I had actually recorded the sound of the hall and audience with a pair of Shure SM81 cardioids with an eye to a possible surround-sound release. Against this reference, therefore, I auditioned literally hundreds of synthesized hall ambiences—something made considerably easier by the Lexicon's well-designed user interface—before settling on an algorithm that sounded like Santa Fe's St. Francis auditorium.

The second key was to not use too much of the synthesized reverberation. In my session-musician days, I used to love listening to the echo-return channels by themselves. These days, I'm more sophisticated. I had the reverberated version of the master available on a second set of Sonic Solutions console faders, and mixed it into the dry signal at what turned out to be a very low level.

So, did I sell my purist audiophile soul for nothing? Check out the sound of Encore and tell me what you think.—John Atkinson