Editor's Choice: Stereophile's Sampler & Test CD Tracks 8-9

[8-9] Beethoven: Piano Sonata No.24 in F-sharp Major, Op.78 (from Complete Sonatas, OrpheumMasters KSP-830)

Performer: Robert Silverman, piano
Recording Venue: The Maestro Foundation Recital Hall, Santa Monica, California
Recording Dates: January 21-24, 2000
Producer: Jim Turner
Piano Technician: Mike Kemper
Sponsors: Aaron Mendelsohn (The Maestro Foundation), David Lemon (OrpheumMasters)
Piano: Bösendorfer 290SE 9' Reproducing Piano
Microphones: two DPA 4011 ½" cardioids (ORTF pair), two DPA 4006 ½" omnis with diffuse-field nosecones (spaced pair)
Mike preamps: two Millennia Media HV-3Bs
A/D Converters: dCS 902D and dCS 904D (both 24-bit) at 88.2kHz
Recorders: Nagra-D (omnis); Tascam DA-38 with PrismSound MR-2024T bit-splitter (cardioids)
88.2kHz/44.1kHz downconversion: dCS 972
Mixer: Sonic Solutions Digital Audio Workstation (4 channels)
24-16-bit Noiseshaping: Meridian 518

The most fundamental decision in location recording is the choice of hall. Given a perfect hall that's the appropriate size for the instrumentation and has a supportive, sympathetic acoustic with the right reverb time for the music, all the engineer has to do is put up a pair of mikes in the perfect spot and roll tape (or, these days, hard drives). But the small concert hall that housed the Bösendorfer Reproducing Piano on which these performances had been prepared was really too small for such a large, powerful instrument. The live sound was bright, the balance thrilling, but the fact that there were walls quite close to the piano added some strong early reflections that at first I found disturbing.

Unfortunately, the Bösendorfer's owner would not allow it to be moved. But as the instrument was automated, I asked the producer to set it playing one of the most dynamic passages in one of the sonatas and keep it in "repeat play" while I did a lot of listening from different places in the hall. I worked out what I was hearing and determined good starting places for the mikes. As coincidence might have it, when the technician came in to work on the piano, he told me that I'd placed the mikes more or less where NPR's engineers had placed theirs when they'd recorded the piano for Performance Today: about 10' in the air and 10' from the piano, looking down the line of the open lid.

I then made test recordings, moving the cardioids about 4" closer at a time on a line toward the piano's soundboard. I thought the position closest in actually revealed the best balance between the sound of the hall and the sound of the piano. However, the mikes then picked up a lot of noise from the piano's action. A piano is actually a percussion instrument: a felt mallet hits a string and produces quite a thump. So I backed off about 3" from that position. This gave me more hall sound than I would have liked, but also lowered that thump to the point where we all felt it was acceptable.

As always, I used a pair of spaced omnis and an ORTF pair of cardioids, recording the four tracks at 88.2kHz. Also as always, the ADC word clocks were linked to allow synchronized playback from my hard drive. My starting point in editing is always to align the tracks in time with one another. To facilitate that, I always have an assistant (or the musician) create a sync pulse when we begin recording, using a slap-stick at center stage. This gives me a time reference similar to the clapper board on a movie set. However, what is the theoretically correct time alignment may not be what is used on the final mixdown.

When I aligned the mikes so they were in perfect coincidence, I was just so aware that the room was too small, too live—unsuitable for such a large, powerful piano. So, for the 16-bit/44.1kHz CD, I ended up ultimately delaying the omnis by several milliseconds, thus smearing over the early reflections of the sounds from the walls. This made the acoustic more anonymous, but also more comfortable-sounding. Yes, this is a move away from purist documentary recording, but one that was justified.

What's interesting about this is that the theoretically correct coincidence of the mikes was acceptable at 24-bit/88.2kHz, because it was so unambiguous. The character of the hall was preserved with such accuracy that you could adapt to it and forget it—just as you would in real life. When I downsampled the recordings to 44.1kHz, I lost a quite a lot of that precise definition of space and time, so the listener could no longer adapt to the sound.

What you should hear: Once you become accustomed to the vivid sound of the big piano in the small room—akin to dropping a big-block Chevy V8 engine into a compact Acura—you can hear how the early reflections help define the recording space without getting in the way of the music. The dynamics are wide, the tonal balance natural, both supporting masterful pianism.

A point to note is that because the piano had been set up for live recitals, it was angled toward the absent audience. The center of the soundstage is therefore the "point" of the instrument, at the treble end of the keyboard. While the bass strings run the entire width of the stage, the short treble strings are actually right of center. This is the opposite of how most piano recordings are made, with the mikes side-on to the instrument, which places the treble strings on the left—as in the Liszt selection on track 16 of this CD. So, no, your channels aren't reversed.