Serenade: the 1996 Santa Fe Chamber Music Festival CD Page 4

One of the things that we couldn't do much about was the wind noise from the clarinets in the Dvorák. When we first started rolling tape, I was astonished to hear what sounded like tape modulation noise---on a live microphone feed! After checking out all the equipment, I went out into the hall and found that it was breath noise. One of the clarinetists, Franklin Cohen, principal clarinet of the Cleveland Orchestra, was apparently having problems with a recalcitrant reed and air was escaping from his lips around the instrument's mouthpiece. We ended up moving the microphones back about 3' for that and other reasons, but it is still noticeable on the second and fourth movements of the Serenade. Hey, it was part of the sound in the hall, and that's what we captured.

Phillips: Do you always record the sound of the empty hall?

Atkinson: I always record a few minutes of silence. I cross-fade to and from it between tracks and between movements. I really don't like the way the CD medium gives you the opportunity to fade to black between pieces of music. To me this is completely unnatural. So on Stereophile recordings, ever since the very first one, after the music has faded away, you have the sound of the quiet hall. The music therefore appears out of the natural ambience of the hall; it doesn't burst forth from a disturbingly dead-silent, digital-black background.

Phillips: Let's talk about some of the other decisions that you made.

Atkinson: After the hall, the next important decision is what microphones to use. There are literally hundreds of different microphones you can use---different-shape pickup patterns, all from different manufacturers---any will be the best mike to use under some circumstance in some halls with some kinds of music, but will not be suitable in others. You then have the decision of how to use those microphones, and there are many competing philosophies of how to use microphones.

Then there's the choice of the other equipment you use---the mike preamp, the cable, the A/D converter, the recorder---all of these will affect the sound one way or another. There's the choice of the editing system you use. If you're using more than a single pair of microphones, there are all the choices you make when you mix-down the multiple microphone signals to just two stereo tracks. And even, it now appears, the mastering process for the CD changes the sound. And until you get that test pressing back from the CD plant, you actually don't know whether you got all those decisions right or not.

Fig.1 Microphone positioning used for this recording.

Phillips: What microphones did you use?

Atkinson: I used the same four microphones this year as in 1995, all from the Danish manufacturer, Brüel & Kjaer [the company is now called DPA]. I use two mikes with an omnidirectional pickup pattern and two with a directional, cardioid pattern. Each mike feeds its own preamplifier, the output of which is recorded on one track of a four-track Nagra-D digital tape recorder.

But with a stereo recording, just as important as the choice of microphones---if not more important---is how to use them. My starting point is this: Theoretically there is really only one way of capturing a stereo image---a sonic picture of where every musician, every soundsource is on a stage---and have it correctly reproduced when it is played back over a pair of loudspeakers. You have to use a coincident technique, ideally with a pair of figure-8 microphones, one mounted vertically above the other and crossed at right angles. (A figure-8 mike captures sound equally to its front and rear but it doesn't capture any sound to its sides.) This configuration is called "Blumlein" miking, as it was first devised by the English engineer Alan Blumlein in the early 1930s.

However, while it's true that when you play back that Blumlein-miked stereo recording over loudspeakers, you'll hear the sound of each musician coming from the right place in space, coincident techniques are generally not so good at capturing the actual sounds, the tonal colors of the instruments. Figure-8 mikes tend not to have full bass response, and depending on which microphones you use from which manufacturers, they won't preserve their theoretical figure-8 pattern at all frequencies. As a result, at some frequencies, particularly in the bass, the image will collapse to mono.

When you listen to live classical music, there's this wonderful feeling of being there. This basically comes from the fact that the sound approaches you from all directions, immersing you in the reverberant field. A stereo recording can't really immerse you in sound coming from all directions, but I feel that it should capture some of that sense of bloom.

So given that the theoretically correct stereo microphone technique is flawed practically, you have to try something else. I start with a variation on a coincident technique using two cardioid rather than figure-8 mikes. I use them in what's called an ORTF configuration, which has the mikes angled about 115 degrees, with their tips spaced about 7" apart. This technique was developed in France and gives a nicely defined soundstage that is a good approximation of a Blumlein-miked stereo image. But as it lacks low-frequency bloom, I also use two omnidirectional microphones spaced quite a long way apart.