RA: Does each instrument have its own phrase database?
EL: Yes. I record musicians playing phrases and capture a wide variety of phrases at different pitches, a collection of phrases that represent the various ways each instrument can be played, especially in orchestral settings. It's a subtle, difficult problem getting the right collection of phrases. I'm working on that now by doing new recordings of musicians and building better databases of phrases.
RA: For a solo violin, for instance, how many notes are actually sampled into Synful Orchestra?
EL: For the current violin database there are 750 notes. Now, one of the interesting by-products of the additive synthesis representation is that it is much smaller than the equivalent recorded PCM phrases—like less than one one-hundredth or even one one-thousandth the size. That's why my entire orchestra fits inside of 32 megabytes of RAM.
RA: Thirty-two megs? That's amazing.
EL: Especially when you consider that many sampled orchestral collections these days are on the order of 300 gigabytes and even into the terabyte range. But it's not just one-to-one that additive is smaller than PCM; it's also that I'm able to reuse a piece of a note over a much wider pitch range. So it's additive synthesis and the way I use this additive synth engine that allows me to transpose over a larger range, and reuse, and recombine materials in a much more flexible way.
RA: All this technology is impressive, but at some point haven't you just thought about giving up and simply investing your time in taking actual violin lessons?
EL: Yes, exactly—that does come up in this line of research! But, of course, we're talking here about the desire to compose and perform realistically on a computer using a keyboard controller. Composers want and need to hear their music realized as realistically as possible but it's still very difficult for them to get performances. When I was a young composer at 18, writing my first orchestral pieces, I would have just loved to have this current technology. I would have even loved to use a traditional sampler way back then, for that matter.
RA: Would you say that emerging computer technology is making a positive difference in the traditional art of compositional music?
EL: Oh, yes—technology is changing the compositional process. It's here to stay. I'm trying to make the technology as expressive as possible.
Breakfast of the music technology titans: Tom Oberheim, Eric Lindemann, David Wessel, Max Mathews, and Keith McMillen dine in Berkeley.
If you've never plugged an expression pedal in to the back of your keyboard, then your performances are just that much less expressive. An expression pedal, which pairs a variable resistor with a seesaw-like surface, can be used to control the performance parameters that each synthesizer engine provides. Want a little modulation on the flute solo or breath in the trumpet part? Just move the pedal up and down as you play and now you're really expressing yourself. The same goes for using a volume pedal or breath controller.
"In order to use Synful Orchestra effectively you need to use a volume or expression pedal or similar controller," says Lindemann. "Synful Orchestra responds to continuous changes in volume or expression with timbre changes. If you're playing a trumpet sound and step on the volume pedal, the trumpet gets brassier. Without this control, you can't contour long notes and the phrasing will sound stiff."
"Synful Orchestra looks at little separations between notes, note overlaps, and velocity values to determine what kind of articulation to use—slurred, tongued, detached, etc," Lindemann says. "So you either need to take care when playing at the keyboard or you need to edit the tracks you've recorded later, in your MIDI sequencer."
Lindemann suggests editing and adjusting the note separations and velocities to achieve the desired phrasing of an orchestral part—a good tip for any MIDI performance. "Synful Orchestra is very responsive," he concludes. "But you still need to control the musical phrasing with MIDI."
These MP3 examples were made by playing the Synful Orchestra with standard MIDI files. The Beethoven string quartet, Copland's "Fanfare for the Common Man," and Hari-Hara were sequenced in Cakewalk Sonar. Stravinsky's Rite of Spring was sequenced in Steinberg Cubase. The production began by inputting basic notes, then adding volume-pedal and velocity adjustments. There is also a bit of pitch-bend in the Beethoven example. Note lengths were adjusted so notes would overlap (producing legato) or separate (producing detached phrasing). All examples feature Lexicon Pantheon reverb processing, and open in new windows when clicked.
The final example is the first of four movements of the ballet Hari-Hara, composed by Lindemann's 18-year-old daughter Anna, the program's alpha tester.
Randy Alberts is an author, musician, and photographer who lives on Lummi Island, Washington.
Return to digitalmedia.oreilly.com