When I first began exploring the world of MIDI in the early 1980s, I was fascinated with this new protocol for interconnecting musical instruments with computer technology. But it wasn’t until 1985 that I became committed to working in this new medium.  At that time, Gary Leuenberger, one of the original programmers for the Yamaha DX-7, the first commercially available digital synthesizer with MIDI ports, called me and a few other musicians to come to his piano store on Market street in San Francisco to listen to what he had created.  Gary programmed the first movement of Bach’s Brandenburg Concerto in D-major into the Yamaha QX-1 MIDI sequencer which was controlling eight DX-7 sound modules (the TX-816 rack).  We sat there eagerly anticipating what we were about to hear while Gary explained what he was doing, and then he pressed PLAY.  Out of the speakers came Bach’s piece and I knew immediately that this medium, in its infancy, was about to evolve and that I knew without doubt I wanted to be a part of it.

I learned to compose music the same way that countless composers learned, composing at the piano, writing scores in pencil, orchestrating at a desk. But by 1985, a new approach had begun to take shape.  Instead of writing music on a score pad, I was playing notes into a sequencer, and when the staff view first appeared in digital audio workstations (DAW) in the early 1990s, I began composing music by popping notes onto a virtual staff using the mouse. 

By 1987 I was creating scores using a software program called SCORE, one of the earliest commercially available music notation programs that produced printed scores at the same level of quality that top publishing houses were producing.  SCORE, however, was cryptic; for example, to change the curvature of a slur, you had to type something like 17 22 11 18 20 6. It was complicated and cumbersome but the output was fantastic. There I was, sequencing and producing electronic music and creating the scores.  No more pencil, nor more music paper.  

But a problem was arising:  As I was adding all the performance markings onto the score--dynamics, articulation marks, hairpins, tempo markings, bowing, phrasing and breathing slurs--I began wondering, why am I writing out all these directions to players if there are no players?   I kept thinking, well, if an ensemble wants to play one of my pieces, I’ll have the score ready and up to modern performance practice standards.  But with every passing year, I was becoming more committed to the virtual orchestra. By the late 1990s, computer processing was getting far more powerful and memory was getting larger.  Sample libraries began to appear.  Instead of a sampled violin consisting of four or five samples on a floppy disc, new libraries were appearing that contained hundreds of samples, using better analog-to-digital converters and including samples of every other note the instrument can play, and in different playing styles.  Multi-sampling was beginning to appear. A multi-sample means that if you play softly you trigger a sample of an instrument playing softly. But if you play harder, another sample of the instrument playing at a different amplitude gets triggered.  So, not only do you hear a change in volume but also a change in timbre (harmonic spectrum), i.e. a trumpet played pp is going to be softer and darker than a trumpet played ff. By the early 2000s, libraries were now appearing with hundreds of thousands of samples, each with two, three and four levels of velocity (multi-sampling). My current library, the Vienna Symphonic Library Cube, consists of around 800,000 samples with every playable note of each instrument sampled in a large of array of playing techniques and styles, recorded in a high-quality environment using 24-bit sampling technology.    

Still, the question lingered:  Why create a score at all if there are no players?  Though its primary purpose is to communicate musical ideas to players, a score also serves to take music “outside of time”, in other words, to symbolically and visually represent musical sound, allowing composers to see and think about how their music is developing and progressing as the composition is created. This gives the distinct advantage of creating more complex music, with many instruments playing different roles in the composition.  Much of the music we know and love could not have been created without a written score. And there’s still another purpose for a musical score: music education.  Studying scores as an adjunct to listening to music gives us insight into how a composer constructed their composition, we can analyze the harmonies and harmonic progressions, we can see how orchestration evolves and changes over the course of a piece.

Since I’m content to be working in the electronic medium, I’m not going to spend valuable time creating instructions for players that don’t exist.  The MIDI sequence already contains all of this information:  phrasing, dynamics, crescendos and decrescendos, articulation of strong and weak beats, the way a note is attacked and released, where a note occurs relative to the beat, tempo changes, meter changes—these elements are all part of the sequencing process that gives music expression, a sense of gesture, intention, and nuance.  MIDI makes a lot possible, but doesn’t make it easy or simple. If one sequences music without programming the details of how a note is played and how a phrase is sculpted, it won’t work as sound.  The precision of computer-based performance is a double-edged sword; without attending to the details of MIDI programming, precision and accuracy can degrade into a lifeless mechanicalness. 

My goal is to continue exploring a medium that, in its own right, continues to hold my interest.   I’ve never been overly-concerned if a MIDI performance sounds exactly like a recording of a real orchestra; if that were my primary intent, I’d write for live players (and I wouldn’t employ synthesizers in my orchestrations).  My feeling is that if you make total realism the goal, you’ll often end up writing in styles that are predictable and too common,  instead of forging ahead and taking artistic risks through the exploration of composition and orchestration that includes software synthesizers and new approaches to form, content, and style.  A MIDI recording can sound musical and sound good without sounding exactly like a recording of an acoustic ensemble. 

I hope this brief statement explains why I continue to write scores and why I continue to avoid the redundancy of putting instructions in the score for non-existent players.  If a conductor wants to play one of my  pieces, I’d add these markings if I believed I would get a good performance and good recording from the collaboration.  In the meanwhile, I’ll continue to work in the medium I love and consider myself fortunate that I can bypass the time-element, politics and economics of trying to get ensembles to play my music.