Key Issues: In Search of the Lost (Harpsi) Chord

You’ve probably figured out that if this had just been a regular harpsichord recording session, it wouldn’t have made the pages of EQ. But we’ll be talking about a very different take on the “classical harpsichord CD,” and a lot of what I learned applies to all kinds of music. Such as. . . .
Publish date:


The harpsichordist, Kathleen McIntosh, has zero attitude, great technique, and is a total workhorse. While a renowned veteran of recording and live performance (she’s appeared as a soloist with the Solistas de La Habana in Havana, Chamber Orchestra Kremlin in Moscow, National Symphony Orchestra of Vietnam in Hanoi, Sinfonica de Santiago in Santiago de Cuba, New American Chamber Orchestra in Spain, Camerata de la Casa del Lago in Mexico City . . . and the list goes on), she was open to experimentation to get the best possible sound.

For example, we started the sessions (at Maricam Studio in Santa Fe, NM) using some classic Telefunken mics and completed all the tracks. But then a stereo pair of Microtech Gefell M-930s (personally tuned by the venerable Jochem Kuhnast) became part of Maricam’s mic locker, and when we checked them out, there was no question that the high end was better suited to harpsichord. Kathleen was willing to re-record all her tracks—with not one word of complaint. These mics had the side benefit of sounding so “right” I needed EQ later on only to solve some specific problems related to the harpsichord itself.


I prefer miking so that the final result sounds like you’re the performer playing the instrument, not someone sitting in the audience. Both Kathleen and Pete Sheehey, the producer, were willing to give it a try. So instead of doing traditional miking, I set up one M-930 close to the performer, pointing toward the higher strings. Due to the length of the harpsichord, I was able to set up the second M-930 toward the bass strings end of the instrument, yet have it far enough away from the “treble” M-930 that there were no phase issues when checking in mono.

The resulting sound was very present, recalling the sound of an FM synthesis plucked string algorithm: Metallic, bright, and defined. But I was concerned that it might sound too synthetic, so we set up two Dirk Brauner Phantom V room mics to provide a “softer” element (Maricam is designed exclusively for recording classical music; the room sound tends to be warm instead of hard due to having Brazilian rosewood panels). The main mics went through Great River MP-2NV Mercenary Edition preamps, and the room mics through Millennia STT-1 preamps.


Maricam’s Scott Irving and I split up the tracking duties. Due to Scott’s background with the Los Angeles Philharmonic (and a Masters in music from the Royal Conservatory in Madrid, Spain), he was comfortable tracking by himself, although about half the time we worked on the project together. He tracked with Digidesign’s Pro Tools HD3 at 96kHz but when I was tracking, I used MOTU’s Digital Performer, again at 96kHz (the studio is Mac-based). Ultimately, though, these DAWs were solely about capturing the sound as cleanly as possible, as I took the audio files back to my studio to edit and mix in Sonar, in large part because of its 64-bit audio engine. Although I remain unconvinced that this matters for most music, with delicate, dynamic acoustic recordings, the 64-bit engine does seem to make a difference.


Harpsichords are noisy instruments, what with string releases and such. But the body of Kathleen’s harpsichord has a wicked resonance at around 50Hz, and because of the close-miking, the main mics dutifully picked it up as a booming, distracting “bump.” Ultimately, I found that the only solution was to “master” each track individually prior to mixing; I used Har-Bal to zero in on these phantom artifacts, and cut them out. As this was mostly below the “real” range of the harpsichord, I could cut pretty drastically without affecting the tone. Furthermore, Har-Bal’s filtering engine has been tweaked a lot in recent versions, so it had no negative effects on the sound. In a way, this process was like cutting out subsonics from other sources; it’s just that the harpsichord’s resonance was a little higher than I would have liked, which made the choice of cut frequencies crucial—the cutoffs were extremely steep.


The DAW editing process was a bit unusual, as the main focus wasn’t fixing mistakes—Kathleen has prodigious, unerring technique, so she’d only do three or so takes. However, the project was based around pieces by Antonio Soler, an 18th century Spanish priest/musician whose pieces recall Domenico Scarlatti but are devilishly difficult and in many places, whimsical. Kathleen loves the pieces, and plays them with a lot of dynamics and feeling; as a result, although different takes could be technically identical, they would often have slightly different “feels” in some sections. I found it almost impossible to choose which individual take was the “best,” so I asked Kathleen if she’d be willing to give up creative control of the editing, and let me just pick which parts I liked best from the various takes and put them together. She agreed, which resulted in the unusual situation of my often replacing a perfectly-played part with another perfectly-played part just because I liked some little millisecond timing difference, or some other element of the “feel.” Generally, these were broad strokes (e.g., replacing the first half of one take with the first half of a different take) in order to maintain a good “flow” to the pieces.

I also encouraged Kathleen to “annotate” her performances by speaking into the mic between takes (e.g., “I really liked the second half on that one”). It was better than taking notes, because her comments were right there on the track. However, I didn’t always agree with her assessments of her playing, so although I trust my artistic judgment I didn’t know if Kathleen would agree. Therefore after editing, I would send her “performance checks” in MP3 format to make sure she was okay with my decisions. Aside from a few small tweaks, she was fine with the edits—which also emphasizes just how helpful it can be for an artist to work with someone who offers a fresh perspective. I then sent a CD with all the tweaked performances to Pete, who signed off on them as well.

Another interesting element was that Kathleen wondered about the possibility of using DAW editing techniques to reduce the level on a few of the release sounds. I suppose that would be heresy in some classical recordings, but she identified spots where the music was very soft, and thought the release sounds had a negative effect on the mood. With traditional room miking this probably wouldn’t have been an issue, but the close- miking brought up the sounds of the release to a louder level than they appeared to be if you were just sitting in the room and listening. So I isolated the release sounds in a few strategic places and brought them down around 4–6dB, which paradoxically, made the harpsichord sound more like “the real thing” that you’d hear in the room.


Using the four mics gave a lot of mixing options. Panning the “bass” mic to the 10 o’clock position and the “treble” mic to the 2 o’clock position gave a nice psycho-acoustic sense of placement; you could almost “see” the harpsichord in front of you. The room mics were panned far left and far right—something I almost never do (I usually pull back a bit from the extremes). But in this case, I could bring up the room mic levels to just where there was a bit of a soft, “pillowy” component to the sound that contrasted well with the ultra-present main mics.

Kathleen wanted an intimate sound rather than a big concert hall effect; I suggested an ambience as if you were in a small chapel, with about 30 people in the room, and that struck her as the best approach. To get some of the hardness of stone walls into the sound, I used a Waves convolution reverb to add just a bit of ambience, but only to the room mics—not the main mics—and panned it to center. This created an ideal soundstage: Harpsichord front and center sitting in a cushiony room sound, and reflecting off the “virtual walls,” a bit of a harder sound to fill in the center “hole” and give more depth.

Basically, I was doing rock miking techniques in a classical context, so I was a bit concerned that it might not sit well with Kathleen and Pete. So, I did two test mixes: One with a full-on, in-your-face, performer orientation, along with a more traditional classical mix with lots of room and a somewhat more genteel sound. I actually thought it was a very good mix in the classical style, and fully expected Kathleen and Pete to choose it. Yet they had absolutely no interest in the “traditional” sound; both vastly preferred the more present sound. Kathleen said it sounded just like what it sounded like to her as she played, which I later found out was a sound she’d always wanted to hear from a harpsichord recording, but never had. Pete was even more emphatic: When I put on the “traditional” mix, he just said “no way!”


With the individual tracks having already been “mastered,” and with the mix doing most of the heavy lifting, mastering was really a question of picking the right order for the 11 sonatas she chose, and doing a few EQ tweaks. In addition to the 50Hz peak mentioned earlier, there was another very strong resonance in the 200Hz range. Although this contributes much of her harpsichord’s “character,” it was emphasized a bit too much due to the close-miking. I again used Har-Bal to bring this down just a couple dB, retaining the character yet making sure it didn’t overshadow the rest of the sound.

The other tweak that surprised me because it worked so well was reducing everything above 15kHz or so. We all wanted a sound that was inherently contradictory: Something brash and present, yet with a smooth, honeyed tone. After experimenting for a bit, I found that cutting off the really high frequencies provided exactly the sound we wanted. I thought about it for a bit, and realized why: The close-miking was picking up all the highs what would normally be lost in a room, if you were sitting back some distance from the harpsichord and there were people soaking up the highs. While part of me chafed at the idea of deliberately cutting the highs, it gave the right sound.


A lot of these concepts apply equally well to recording any solo artist. In fact, I plan to incorporate some of those techniques in two upcoming projects at Maricam, including a project with Kathleen and flautist Eugenia Zuckerman. But flute is a whole other story!