The Well of Fancy Dry: a composition for solo flute and live computer processing
Instruments and music in our time: general background
With the acceptance of the phonograph near the end of the 19th century, our expectations and understanding of musical sounds and experiences began to change. Today, that transformation is complete. Whereas a previous age imagined music in terms of “real” instruments (acoustic, mechanical devices like pianos or violins) and concert halls, most of the music we hear today can be described as electronic—or at least requiring electronics. We are immersed in an acoustic environment of sounds created through electronic means, only some of which are musically useful. These sounds are omnipresent: computerized beeps at the ATM, automated phone systems with synthesized voices, as well the musical instruments that won’t work during brownouts. But to an even greater extent, our music, even when played on “old fashioned” instruments, is electronic because we experience it most often in recorded form: captured using microphones, stored on mechanical, magnetic or digital media, and experienced as played back on loudspeakers (usually in a time and place different from its creation). Music is everywhere now—piped into the doctor’s waiting rooms, on your alarm clock when you awake, even carried along with you while jogging. If all this now seems completely natural, it's because of a slow evolution. We now live, to paraphrase Walter Benjamin, in the age of musical reproduction.
Musical background and an aesthetic position
What we now accept as music has also changed as our means to produce and to experience it have been technologized. In popular forms for example, phonographs and samplers have been instrumental in the development of music such as rap, which uses previously recorded music as a source for new “instruments.” The living tradition of classical music has also changed with the times. Concert music has always been interested in making use of the latest technology. The modern piano was developed in Beethoven’s time, and his piano music is only conceivable in light of the invention of a more dynamically responsive instrument. The modern history of electronics in classical music begins at the turn of the twentieth century with the futurists and composers such as Honneger, Antheil, Varese and Russolo who included the sounds of the modern age in their concert works—sirens, the noise of airplanes or trains, and invented “intonarumori” (noise intoners). Decisively modern perhaps, was the introduction of the magnetic tape recorder as a musical instrument by Pierre Schaffer just after World War Two, and the development of a musical style, musique concrete, composed exclusively of recorded sounds. These are all the precedents for my own compositional work.
Trained as a classical composer, I use both new and old instruments in my music: acoustic instruments such as flute, clarinet, cello and percussion instruments, alone and in combination with live, interactive electronics. I am both extending the goals and traditions of the classical traditions, as well as making creative use of new technology.
My composition, The Well of Fancy Dry, for solo flute with live computer processing (described below) is part of a series of works, each for a single acoustic instrument in conjunction with a specially designed piece of computer software. This series explores the range of possible ways a live musician might interact with electronics, sometimes as accompaniment, at times as musical leader, and all the combinations in between.
Tools and equipment: using a computer for music
The tools of computerized sound production have become ubiquitous, almost as much so as music made with computers For example, consider the powerful software now available to “rip” and encode music in mp3 format. As with the Internet more generally, the formerly inaccessible world of codecs, digital bit transfer rates, and file formats is now becoming widely available to home hobbyists. Fifteen years ago, music made with computers was the exclusive domain of music and computer science departments at large research universities such as MIT and Stanford. With the advent of powerful personal computers and the Internet, much of this work is now possible with more modest means. Today, computers can be used to make music in a wide variety of ways. Software “sequencers” can be used to record and play back the performance of dedicated synthesizers and keyboards using MIDI (Musical Instrument Digital Interface), acting not unlike a digital player piano roll. Another category of software represents sound directly as a stream of numbers (digital audio, as in a compact disc), and can be used to generate (or synthesize) completely new sounds, or to alter (edit and process) existing acoustic sounds.
I created the software for The Well of Fancy Dry in Max/MSP, an "erector set" of sorts for audio, MIDI and multimedia. Originally developed at the French national computer music studio, IRCAM, Max/MSP provides an intuitive and versatile way to program real-time digital signal processing (DSP) applications. This environment offers at least two artistic advantages: it allows one to design and use unconventional DSP algorithms not readily available from commercial audio effects processors, and it allows a single program to produce many different musical results, dependent on the nature of the input or on decisions made during performance by the computer or by a performer.
Figure 1 shows an example of simple program “written” in the graphic language of Max/MSP. In this signal network, MIDI information coming from a digital keyboard (connected to the computer) controls the audio generating portion of the program. A simple frequency modulator (FM) synthesizer can be played by an external keyboard which provides a trigger (on/off), pitch (frequency), and volume (amplitude envelope) information. Functions, such as system level MIDI input (notein), a sine wave lookup table oscillator (cycle~), and arithmetic functions (+~ and *~) are represented by the rectangular boxes (called “objects”). Control signals (MIDI number streams) travel through the solid black lines. Audio rate signals pass through the fuzzy yellow/black pipe cleaner lines.
User interface objects such as sliders, audio meters, presets, buttons, and switches provide a means to monitor and control the software during performance. Figure 2 shows the user interface windows for the The Well of Fancy Dry performance software. The largest window (with the title at the top) provides access to the composition’s common controls: DSP algorithm selection for each movement, audio levels for individual processes, input/output monitoring, fader presets, and setup/testing facilities. The window in the lower left corner provides feedback on the input signal, while there is a separate window containing the processing algorithms for each of the three movements. The first movement is shown in figure 2. Further windows with the “guts” of the DSP programming are nested within the object boxes shown in this window.
Algorithms and composition
In The Well of Fancy Dry, a human performer plays a flute from a conventionally notated (though extended) musical score (Figure 3). The sound of the instrument is fed to the computer (an Apple Macintosh PowerPC) through a microphone, then to a digital–to–analog convertor (DAC) where it is accessible to the software. The software both processes the instrumental sound and uses the flute to trigger newly generated synthetic sounds. The output of the computer software is then amplified through loudspeakers on stage with the flutist. The computer responds quickly enough so that it may accompany the flute as it plays with only a few milliseconds delay between the input and output of the sound.
Each of the three movements in the composition involves different types of processing, as determined by the musical intent and need. In the first movement, the software analyzes various aspects of the flute’s performance: the pitch played (divided into high, middle, and low registers), the dynamic level (loud and soft), and percussive quality of the attack. Each of these characteristics triggers a different response from the software, such as ring modulation of the source, spatial placement (along signal defined panning paths) and the degree of reverberation. The effect is a multi-layered “super flute” which makes literal the polyphonic (multi-voiced) implications of the compositional writing.
In the second movement, the software monitors the flute’s performance to determine phrasing (breaks between musical ideas). Each complete phrase is recorded into its own semi-permanant location (ten independent audio buffers, circularly filled) processed and played back algorithmically when triggered by subsequent flute playing. As an example, figure 4 shows the programming that controls the recording of phrases into numbered audio buffers. The first of these full buffers is shown at the bottom, displaying the flute sound as a time domain wave form. The effect is of a gentle melodic line in the flute which provokes distant, obscured memories.
In the final movement, a pitch tracker again calculates the fundamental (or perceived) frequency of the flute. Each note of the flute triggers a series of synthesized responses. Cumulatively, the computer is performing on a synthesizer built from a resonance model of a tubular bell. When the flute generates enough notes quickly enough, the full timbre of the bell sound is evident, otherwise, only portions (an incomplete set of partials) are played at any given time. The cumulative effect is an almost Gamelan-like texture.