musicL
Andrea J. Stein, M.P.S.
Master’s Recipient: Interactive Telecommunications
Program
New York University, New York, USA
ajs327@nyu.edu
Abstract
musicL is a real-time system for the
generation and synthesis of multi-voiced compositions. Employing a set
of generative algorithms, musicL disrupts our traditional expectations of what
music does.
musicL is a generative system for simultaneously
creating music and its notation. The generative properties of musicL are
determined by L-System equations (a mathematical theory developed to describe
the branching and growth of trees). L-Systems are literally instruction sets
that, among other things, dictate when to go forward, left, or right, by how
many steps, and at what angle (see Technical Specifications). Therefore,
the rules that govern musicL are a set of spatial directions which manifest
sonic and visual characteristics. In other words, each unique L-System
simultaneously draws a pattern and triggers sound.
In its present form, musicL allows for the arrangement of both
pre-defined and random L-System compositions. The interface consists of the
Elements, Stage, References, and Toolbar (fig 1).
Each Element is essentially a specific L-System. They are displayed as
thumbnails which illustrate their emerging patterns. There are currently 9
pre-defined Elements, as well as one random Element. The random Element is
represented by r and assembles both its instruction set and its values
randomly each time it is used. Consequently, each occurrence of the random
Element is theoretically unique.
The spatial coordinates of the Stage are mapped to pitch (top to
bottom) and amplitude (left to right). Each Element is dragged on to the Stage
to initiate its movement and sound. The Element’s vertical range of motion (and
thus its pitch range) is limited to 100 pixels above and below it’s release
point on the Stage. A maximum of ten Elements, in any combination, can exist on
the Stage at any given time. Thus, ten occurrences of one Element is allowed,
as long as the total number is not exceeded.
Figure 1: The
Interface
The References (displayed in a row) indicate which Elements are
currently active on the Stage. As an Element is activated, its thumbnail will
appear in the “Reference Row”. Additionally, an Element can be inactivated
(removed from the Stage) by clicking on its Reference.
Finally, the Toolbar contains musicL’s visual options. The patterns
which develop on the Stage can be viewed differently, depending on the mode set
in the Toolbar (fig 2). The Toolbar also contains a ‘pause’ button, allowing
the user to observe the visual characteristics of musicL in a static form.
Figure 2a:
Demonstrating ‘Full’ Drawing Mode
Figure 2b:
Demonstrating ‘Connect’ Drawing Mode
Generative Forms
Generative principles, as they
apply to music and composition, are understood more clearly if discussed in
contrast with classical composition techniques. Classical compositions are
written with a pre-defined, specific, and desired outcome. The directions are
precise and inflexible. In the end, the classical composition belongs to its
composer. However, generative compositions by nature eliminate
the idea of single authorship. Relinquishing control over the final product,
the generative composer allows unexpected and emergent properties to arise.
Brian Eno clarifies the distinction as he states “classical music … specifies
an entity in advance and then builds it. Generative music … specifies a set of
rules and then lets them [the rules] make the thing.”
"...in a sense it
means 'putting things together', and I put them together in such a way that
they're not fixed in a fixed way but flexibly work together."
-John
Cage, on what it means to be a composer.
In the early 50’s experimental
composers such as John Cage, Earle Brown, Christian Wolff and Morton Feldman,
gained notoriety relinquishing control over the execution of their art. Their
motivation stemmed from “the immediate desire to deal with what sound is,
rather than what the composer may think it is or decides he wants it to be.”
John Cage tackled this problem by introducing chance and random procedures into
his work. Feldman denounced methodology in favor of instinct. Brown’s interest
in artists such as Calder and Pollock accounts for the spontaneity and
open-form mobility found in his music. By providing performers with vague and
often incomplete instruction sets, emphasis was, shifted away from the
composer, and the individual performer was given more control.
Notation
Staff notation, used since at
least the 17th century, with its origins dating back to the 9th
or 10th century, is a very precise tool for dictation. The New York
School’s unique methods, however, no longer fit the traditional framework of
composition, and often called for new methods of notation. According to Cage,
dictation does not have to be built into the notion of a score. “…if there are
several parts [of a score] and there’s no fixed relationship, then there’s
nothing built-in. No fixed relation built in.”
Morton Feldman was the first to
use non-representational graphic notation. He divided the pitch range of each
instrument into high, middle, and low, and represented each range as a
rectangle on graph paper. The specific pitch and boundaries were left as
decisions to be made by the performer. Christian Wolff composed one piece by
writing notes vertically down the page, but had the performers read and play
the piece from left to right. Cage consulted the I-Ching for note
placement, while leaving rhythm and detail to the performer. Brown’s famous December
1952, a musical analogue to Alexander Calder’s mobiles, was notated by an
abstract series of floating rectangles.
musicL speaks to this tradition of
unique forms of musical notation. Here, aspects of notation are separated into
two parts:
1)
the
L-system algorithm, which gives impetus to the music, and
2)
the
visual/graphical representation of the music, which leaves artifact or record
of what has transpired.
musicL was created in Java and
runs as an applet in a standard web browser. The sounds are synthesized in
real-time, using a Java API called JSyn (written by Phil Burke). The properties
of musicL are derived from generative algorithms based on L-Systems.
L-Systems
An L-System is a mathematical theory
that describes plant development. It is represented by a series of symbols
representing specific commands. The central concept behind L-Systems is the
notion of rewriting. The Algorithmic Beauty of Plants describes rewriting as “a
technique for defining complex objects by successively replacing parts of a
simple initial object using a set of rewriting rules or productions.” All
L-Systems consist of an axiom and a rule. The axiom specifies an initial
condition and the rule dictates what part is to be rewritten, and by what. The
depth indicates the number of times this replacement will take place.
musicL has been commissioned by to
provide the score for upcoming performances with the Esse Aficionado Dance
Troupe at the Merce Cunningham Dance Studio in New York City.
musicL was originally presented to the Chair and Faculty of New York University’s Interactive Telecommunications Program in May, 2003.