Computer Music Languages, Kyma, and the Future

Author(s): Carla Scaletti

Source: Computer Music Journal , Winter, 2002, Vol. 26, No. 4, Languages and
Environments for Computer Music (Winter, 2002), pp. 69-82
Published by: The MIT Press

Stable URL: https://www.jstor.org/stable/3681771

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms

The MIT Press is collaborating with JSTOR to digitize, preserve and extend access to Computer
Music Journal

This content downloaded from

132.248.9.8 on Tue, 28 Mar 2023 15:55:33 UTC
All use subject to https://about.jstor.org/terms
Carla Scaletti
Symbolic Sound Corporation
Computer Music
P.O. Box 2549
Champaign, Illinois 61825-2549 USA Languages, Kyma,
[email protected]
www.symbolicsound.com
and the Future

Eric Lyon organized this symposium around the they have had a longer-lasting and deeper influence
following two questions. Why has some computer on the evolution of music.
music software survived and developed a follow- This article is organized into three sections. The
ing? Where is computer music software today, and first section is an identification and discussion of
where might it be headed in the future? Acknowl- factors that can contribute to the success and lon-
edging that the term "computer music" is by now gevity of a computer music language. In the middle
redundant, since virtually all music involves the section, I try to illustrate some of those factors
use of computers in some form or another, he (like extensibility) using specific examples from the
posed these questions with respect to computer Kyma language. The last section is a speculation on
music software that supports experimental music. the role computer music languages could play in a
Experimental music is not limited to any specific future world where art, the economy, and human
musical style or genre. An experimental musician beings are very different from the way they are
is one who approaches each act of musical creation today.
in a spirit of exploration and innovation, often with
the goal of inventing new kinds of music that have
never been heard before. Factors Contributing to the Success
I would like to take Eric Lyon's refinement a step of a Language
further by observing that all of the software exam-
ples included in this symposium (along with some
Why have some computer music languages sur-
others that are not represented here) belong to a
vived over the years and attracted a sizeable num-
special category of computer music software called
ber of users? Although we would like to believe
computer music languages. Most software packages
that the only reasons for the success of these lan-
can be classified as utilities; they perform a well-
guages are their inherent attributes, many of the
defined, familiar function that is needed by a large
factors contributing to success have little to do
number of people. A software package that emu-
lates all the functions of the traditional multi-track
with the language or technology itself. There are,
for example, the lucky accidents of geography,
recording studio would be one example of a utility.
economic/social class, and chromosomal makeup.
But the pieces of software that Eric Lyon has cho-
Given these "accidents" of birth and fortune, what
sen to include in this symposium are different;
other necessary (but not sufficient) factors can con-
they are examples of computer music languages.
tribute to the longevity and acceptance of a com-
A language provides one with a finite set of
puter music language?
"words" and a "grammar" for combining these
words into phrases, sentences, and paragraphs to
express an infinite variety of ideas. A language does
not do anything on its own; one uses a language to Reasons Intrinsic to the Language
express one's own thoughts and ideas. This is what
makes these particular software packages so open,Several factors intrinsic to a language contribute to
extensible, and useable in ways unanticipated by its relative success. A language is successful if peo-
their authors. And that is why, although they mayple are using it successfully. It does not matter how
never command the same market share as utilities, elegant or beautiful a language is in theory or on
paper if no one is using it. Any language that has
longevity and a following also has a corresponding
Computer Music Journal, 26:4, pp. 69-82, Winter 2002 list of creative projects that have been successfully
C 2002 Massachusetts Institute of Technology. completed using that language.

Scaletti 69

This content downloaded from

132.248.9.8 on Tue, 28 Mar 2023 15:55:33 UTC
All use subject to https://about.jstor.org/terms
 A language is successful if it answers a need that isfy several of the issues at once. Even more
is not otherwise satisfied. How does a language important is the task of giving people what they do
gain users in the first place? First it must ade- not yet know that they need. People do not always
quately address some basic needs, and then it must know that they want something if they have never
also answer some need (tangible or otherwise) that seen it before. But that is often exactly what they
is not currently being satisfied elsewhere. want and need most of all: something new.
A language is successful if it is able to express A successful software designer must maintain a
the unanticipated. In a computer music language, balance between responding to user feedback and
just as in a natural language, it should be possible maintaining a consistent vision for the language.
to say something new. For example, a highly re- At times, it is the users who are specifying the lan-
strictive data-flow editor-while it may protect one guage and at others, it is the author who forges
from mistakes-also prevents one from discovering ahead into new territory. This is not so different
the sounds that its author could not imagine. A from a composer who gives the audience some fa-
successful music language does not impose a par- miliar material while also nudging them along into
ticular musical style upon its users; it must be flex- new musical territories.
ible enough and abstract enough to generate
musical utterances in any style and in any genre.
A language is successful if its underlying data Reasons Extrinsic to the Language
structure can support extensions and multiple in-
terpretations without violating the original model. A language is successful if one can learn it and
This ability to express the unanticipated extends learn from it. Whereas the success of a utility de-
beyond the users of the language to the developers pends in large part on its being completely obvious
of the language. The underlying data structures of aand easy to use without having to refer to the man-
successful language should make it easily extensi- ual, computer music languages, by definition, re-
ble, even in directions that the designer had not quire more of their users. Simply knowing the
originally anticipated. When a language has this syntax of a language does not automatically give
characteristic, it can evolve and adapt. one something interesting to say. The most suc-
A language is successful if its author can strike a
cessful users of computer music languages are
balance between providing users with what they those who also take an interest in learning more
say they need and what they do not yet know that about sound, music, and structure. For a computer
they need. It comes as no surprise that when the music language to be successful, there must also be
author of a computer music language is willing to opportunities for learning more about sound and
make adjustments and additions to the language in music creation in the context of that language:
response to user feedback that those languages be- workshops, courses, tutorials, extensive documen-
come more successful and develop a loyal follow- tation (via print and other media), books, lectures,
ing. But the feedback process is not quite as directan online forum, technical support, etc.
as it might seem at first glance. If an author simply One of the beneficial side effects to learning a
went through the "wish lists," item by item, add- computer music language is that the basic princi-
ing each feature that was asked for, the result ples are completely transferable to other languages
would be a bloated and jumbled mass of features and even to other utilities and hardware setups.
without the logic or framework necessary for the Learning a computer music language provides one
user to make sense of it all. with a basic understanding of sound, logic, and
Instead, the authors of the successful languages structure that one can apply no matter how tech-
have had to digest and analyze the feedback first nology evolves in the future. By contrast, the
before making modifications to the language. They knowledge of which buttons to press in a special-
have had to combine all the specific requests, gen- purpose utility becomes obsolete as soon as that
eralize them, and produce meta-solutions that sat- utility is upgraded or replaced.

70 Computer Music Journal

This content downloaded from

132.248.9.8 on Tue, 28 Mar 2023 15:55:33 UTC
All use subject to https://about.jstor.org/terms
 A language is successful when it has a commu- step towards creating a successful language is to
nity of users. Although an artist is almost by defi- start writing music software in the first place. Cre-
nition independent, there is also some stimulation ative environments where many people are writing
and practical assistance to be derived from interact- software can remove the barrier of "over-
ing with other people who use the same language. reverence" that sometimes surrounds existing lan-
Successful languages have groups of users who pro- guages (which may appear as if they had sprung
vide technical support, new developments, educa- fully formed out of nothing and to have existed for-
tion, and appreciation for each other. ever). A healthy environment for language develop-
A language is successful when it serves as a ment is one in which there is a lot of software
nexus for interdisciplinary cross-fertilization. The experimentation and where the conversation is just
most powerful function of any language (whether it as likely to turn to software architecture as it is to
is a programming language, a computer music lan- the weather or the news.
guage, or a natural language) is to act as a nexus for A language is successful when people are ready
the cross-pollination of ideas among different for it. In technology, it seems that if an idea arrives
fields. Each person who uses a language also has even a little bit too early, it has to struggle for sur-
some effect on its development. When a language is vival, but once the time is right, there is an explo-
general enough to satisfy the needs of users having sion of development and growth. People have to be
different backgrounds, different goals, and different conceptually prepared to receive and understand
knowledge sets, it can also act as the vector for what is presented to them, and the essential tech-
transferring ideas and protocols from one discipline nological infrastructure must be in place in order to
into another. When the ideas and protocols are ap- reach potential users and achieve the basic func-
plied in the new field, they acquire additions and
tionality.
alterations that in turn find their way back into the
A language is successful when people say that it
language, and so on.
is successful. If you live in a village where every-
A language is successful when its author uses it
one believes in witchcraft and the local witch puts
regularly. Using one's own software gives one a dif-
a public hex on you, it won't matter that you are a
ferent perspective than is possible from merely
cheerful person in the best of health; everyone will
writing it. There is no better way to fully identify
start treating you as if you were already dead. What
with the users of the software and get ideas for im-
is said becomes the truth. And when something is
provements than to compose with the language,
perform on stage with it, and otherwise use it in in print, it carries even more authority. So having
the same way its users do. The more the author some good things said about the language in the
can do this, the better the language becomes. media, on the web, in journals, and in the history
A language is successful when the people behind books lends the language an air of credibility and
permanence.
it are committed to its success. In the end, the sin-
gle most important factor in the success of a lan- A language is sometimes successful due in part
to pure luck. The successful languages are the one
guage is that its author, its users, a company, and/
or an institution are committed to its continued whose designers were prepared to recognize and
existence. Behind every successful language, one take advantage of being in the right place at the
can find an individual or individuals dedicated to right time.
the perpetuation and further development of that A language is successful if it has contributed
language. Some very interesting and beautiful lan- ideas and stimulated new developments in the
guages have had to be abandoned by authors who field. Even the most successful of languages will
felt compelled (economically, professionally, or not stay around forever, but the ideas from those
simply due to their own breadth of interests) to languages will endure and influence the direction
pursue other projects. of future developments. Human beings build lan-
A language cannot be successful unless it first guages (natural languages, computer languages, and
exists. Although this may seem obvious, the first music) as a collective project, often with the same

Scaletti 71

This content downloaded from

132.248.9.8 on Tue, 28 Mar 2023 15:55:33 UTC
All use subject to https://about.jstor.org/terms
blind dedication displayed by termites building a provide no real-time interaction. To my mind,
nest. Each individual works, almost by instinct, on though, these shortcomings were overshadowed by
some small portion of the project without being the promise of the structural, organizational, and
able to fully conceive of the part it plays in the algorithmic power of the computer, so Kyma was
larger structure. The higher-level structure (in this also inspired by these earlier computer music lan-
case "human knowledge" or "culture") is an emer- guages.
gent property of the myriad of smaller structures Kyma also drew inspiration from a language
built by individuals. called Smalltalk and the paradigm of object-
oriented programming. To me, "object" suggested
something like the snippet of audiotape that I
could hold in my hands. But it also suggested a
The Kyma Language
more abstract conceptual grouping of atomic or
compound sound objects that could be manipulated
Origins or viewed as a single entity and then "zoomed" to
reveal arbitrary levels of detail. Kyma was also in-
fluenced by the courses I took on data structures,
Where did Kyma come from? Kyma comes in part
mathematical logic, automata theory, discrete
from a fusion of my childhood interests in both
mathematics, and graph theory at the University of
music and science and from pencil-and-paper at-
Illinois. In a 1986 course on Programming Lan-
tempts to algorithmically map patterns in nature to
guage Principles, I wrote a term paper called
patterns in music written for acoustic instruments.
"KYMA: A Computer Language for the Representa-
In part it arose out of the atmosphere at the Uni-
tion of Music" describing an object-oriented pro-
versity of Illinois in the early 1980s, an environ-
gramming language for analog circuit simulation,
ment rich in living examples of composers and
score representation, rule-based composition, and
engineers who were creating their own computer
direct manipulation of waveforms.
music languages and instruments or extending ex-
I started writing code for the first version of
isting languages to suit their compositional require-
Kyma in Apple Smalltalk running on an Apple
ments. (Consider, for example, Herbert Brfin's Macintosh 512K in 1986. In 1987, I modified Kyma
elegant and algebraic SAWDUST language, Salva- to make use of the Platypus signal processor as an
tore Martirano's SalMar language/instrument for accelerator and demonstrated it at the 1987 Inter-
real-time composition, John Melby's extensive national Computer Music Conference in Cham-
score manipulation subroutines, Jim Beauchamp's paign (Scaletti 1987). In 1989, Kurt Hebel and I had
voltage-controlled analog synthesizer and his later concluded that it was not in keeping with the man-
forays into hybrid synthesizers and computer mu- date of a university research laboratory to build,
sic languages, Sever Tipei's MP1 algorithmic com- distribute, and support hardware and software. So
position language, and the CERL Sound Group's we formed a company, running it for the first three
long line of digital synthesizers and music software years from our third-floor student apartment. We
culminating in the Platypus user-microcodable assembled hardware in the kitchen, developed soft-
DSP designed by Lippold Haken and Kurt Hebel in ware in the spare bedroom, used a closet stuffed
1983.) In part it arose out of my fascination with with sound-absorbing blankets as our recording stu-
tape music-with being able to literally hold bits of dio, and used the living room for faxing, mailing,
sound in my hands and to manipulate time by rear- packing, and shipping. That first summer, we could
ranging the pieces and taping them back together not afford to run the apartment's air conditioner,
like a filmmaker. And in part it arose out of my but we continued programming and hardware-
frustration with computer music languages of the designing into the fall and winter, finally hearing
time, which felt almost like a step backward from the first sounds from the Capybara at midnight on
tape music, because they were defined in terms of New Years's Eve (in the first few minutes of 1990).
music notation played on "instruments" and could Thanks to the extraordinary vision of our first cus-

72 Computer Music Journal

This content downloaded from

132.248.9.8 on Tue, 28 Mar 2023 15:55:33 UTC
All use subject to https://about.jstor.org/terms
tomers, who understood what we were doing long The definition is recursive, so one can build arbi-
before anyone else took us seriously, we managed trarily long chains of functions of other functions
to bootstrap Symbolic Sound Corporation and (for example a LowPassFilter of a HighPassFilter of
move into our first real office in 1992. a Mixer of three mixers and the microphone input).
Since that first version of Kyma in 1986, there Some of the functions are temporal functions
have been five major software revisions, a port called TimeOffsets and SetDurations. Using these
from the MacOS to Windows in 1992, and ports totemporal functions within a structure, one can cre-
five different hardware accelerators. The user base ate sequences of Sounds, mixes of overlapping
has expanded from a single composer/software de- Sounds, and other time-varying structures.
veloper working in a university research lab to an One of the ramifications of this abstract, recur-
international community of musicians, sound de-sive definition is that Kyma makes no distinction
signers, and researchers. Symbolic Sound has been, between "samples," "live audio input," and syn-
for us, an alternative means to the end of continu- thetically generated signals. They are all Sounds
ing our research and teaching. Our research and that act as sources, and they can be manipulated
ideas are embodied in software, hardware, and and composed using the same sets of unary and
sound, and we have been privileged to teach (and n-ary functions. Another ramification of the Sound
learn from) heterogeneous groups of individuals of object definition is that Kyma does not draw a
every age and background, many of whom have distinction between "instrument" and "score."
also become our colleagues and friends. Instead, it provides an abstract way to build hierar-
In presenting this concise personal history, I chical structures that might or might not corre-
hope to highlight the fact that no computer lan- spond to traditional musical organization.
guage ever springs fully formed out of nothing. This recursive definition forms the basis of
Each language has a distinctive flavor to it im- Kyma. Kyma would still be Kyma even if it had no
parted by the primordial "soup" in which it was graphical user interface and no hardware accelera-
spawned-a history, a personality, and a reason for
tor (and in fact the first version of Kyma was com-
being. Each language is the complex embodimentpletely text-based and ran on a single processor).
of a theory of sound, music making, and structure.
The basic assumptions of each language show the
imprint of the designer's personality, experiences, Multiple Viewpoints on the Same Abstract
theories, and training. Data Structure

Rather than calling Kyma a "graphical language,"

Definition and Ramifications of the Sound Objectit would be more accurate to call it a language that
provides multiple ways of viewing and manipulat-
Kyma is a language for specifying, manipulating ing data. The current graphical representation of
and combining sounds (Scaletti 1997), and it is Kyma Sounds has evolved over several years as I
based on the following definition: have tried to discover the most direct and appropri
ate representation for understanding and manipu-
A Sound is defined to be a Sound S, a unary
lating the Sound structures. The abstract structure
function of another Sound f(S), or an n-ary func-
came first, and the graphics evolved (and continue
tion of two or more Sounds f(sl, s2, ..., sn).
to evolve) in order to elucidate the structure.
For example, a Sound might be a source of sound
(like the audio input, a sample, or a noise genera-
Evolution of the Graphic Sound Editor
tor). It could be a unary function of another Sound
(like a LowPassFilter of another Sound). It could Initially, a Sound was specified symbolically as
also be an n-ary combination of several Sounds Smalltalk code. This quickly evolved into a kind o
(like a Mixer with twelve input Sounds). "selection-from-lists" interface (shown in Figure 1)

Scaletti 73

This content downloaded from

132.248.9.8 on Tue, 28 Mar 2023 15:55:33 UTC
All use subject to https://about.jstor.org/terms
Figure 1. Kyma screenshot automatically pasted intowas displayed in the lower
from 1987. After selecting the center pane. The hier- left pane. The panes to the
a Sound class from the list archical structure of the left and right of the center
in the upper left and a Sound was represented us- pane are lists of previously
creation method from the ing indentation levels created Sounds that could
list at the upper right, the (lower right), and the cur- be used as inputs to the
creation code would be rently selected subSound Sound in the center pane.

FunstioT, I:mT~l tcITi lT-i;- .. .........r....rrnsP~?1`. r?,is :"~~~:;~iso !,:,: W-YVQP%,*,

. . . .. . . . . . ....,. .. .
Fvoiett ? 14t..6.Ifr m .. .. .................. ...................

........ . . ..... . , ... . . . , , ,

. . .,. ...
P....u , i. A * 1,,4
. ..k ...... .... . .

..... .. . . . .. . . . . . . .. . .. . . . . . . . .. +i
_ . . . . . .. .. . . . .. . . : . ... . . . . . ... . . _ . ........ . ~ i + .; : . ..i.
? . . .. ..: .. . ? .. . 1..: 1.1.i . l .W,. u+ ~ ..+ . d,~ b..+ + s . .. .

.I........ ... . . Wix~ar:1143.

h .. i .1
S.. ' . S b
.FrrA.
awr*n.4W ,6441 .... "N T7
.. t
Vea n:.
Fit
Fr410

in which one could select a class and its creation been right at my fingertips (literally) all along.
method from lists, and they would then be auto- Every time I tried to describe the structure of a
matically typed into a window. Previously created Kyma Sound, I always drew a picture like the one
Sounds appeared in lists on either side of the codein Figure 3.
window, so they could be used as arguments to All this time, I had been describing Sounds in
new Sounds. Its indentation level represented the one way but representing them on the computer
position of a Sound in the hierarchical structure, screen in a different way. It finally occurred to me
and one could select any level and listen to it atthat a drawing of the Sound structure could be
that point. more than just a tool for describing and under-
In an attempt to further reduce the need for typ- standing the Sounds: it might also be the most di-
ing, I changed the interface to the "Russian doll" rect way for people to create and manipulate the
style interface shown in Figure 2, where Sounds are Sounds. In other words, the same representation
represented as "containers" of other Sounds. could be used for both analysis (i.e., a description
Double-clicking on a Sound opened a window after the fact) and synthesis (i.e., creating and modi-
showing the Sound's input(s) and parameter values, fying the structures). For me, at least, this was an
double clicking on those inputs revealed their in- important conceptual breakthrough. It was the first
puts, and so on. time I realized that synthesis is just one specific
This succeeded in revealing the recursive nature case of analysis.
of the structure but was not very good at showing I changed the graphical representation of Sounds
the overall structure and the connections between from boxes-within-boxes to something that looked
the Sounds. Finally, I realized something that had like the screen shot in Figure 4, a graphical repre-

74 Computer Music Journal

This content downloaded from

132.248.9.8 on Tue, 28 Mar 2023 15:55:33 UTC
All use subject to https://about.jstor.org/terms
Figure 2. Hierarchical Figure 3. Drawing of a
Sound structure repre- Kyma Sound structure
sented as boxes within (Scaletti 1989).
boxes (Scaletti and
Johnson 1988).

+
Mixer

Name twoVoices
Scale SubSounds
1.0

0.6

-0.2

-0,n

Name [U]

"[U] duration inSeconds" 2,0

samples

"[U] frequency inHertz" 207

pitch

1.0 "[U] formantl "[U] formant2 1,0

06 inHertz" 2200 pitch inHertz" 2650 pitch

Sp"[U] bandwidthl
inHertz" i "[U] bandwidth2 pc
70 pitch inHertz" 45 pitch
play replay help save replace fullReplace

sentation of the functions, clearly influenced by

the representation of trees and Directed Acyclic
Graphs (DAGs) in computer science. The default
signal flow direction was from bottom to top (an
influence of the textbook illustrations of Music N concatenation

unit generators).
The direction of signal flow was user-selectable,
so one could reverse the signal flow direction or Delayed

rotate it 90 degrees. Influenced by DSP textbooks

and papers, I eventually decided to standardize
DurationTransform
the Sound representation to one showing the sig-
nal flowing from left to right (as shown later in
Figure 6).
FrequencyTransform FrequencyTransform
Note that through all of these changes in graphi-
cal representation, nothing about the underlying
Sound structure changed. However, even some- Pluck
thing as simple as changing the direction of signal

Scaletti 75

This content downloaded from

132.248.9.8 on Tue, 28 Mar 2023 15:55:33 UTC
All use subject to https://about.jstor.org/terms
Figure 4. Early GUI
version of the Kyma Sound
structure (Scaletti and
Hebel 1991).

parameterlzedPluck

(dwatio (- ((./ ....rt )its

a-,

flow had a noticeable effect on the way people un- I also noticed that whenever I wanted to create a
derstood and manipulated that underlying structure. parameter control function, I would first draw the
desired control function on paper and then work
Evolution of the Graphical Timeline out the arithmetic combination of functions to
Kyma's Timeline editor is yet another view of that generate that shape. That is why the Kyma.5 time-
same underlying data structure. Kyma Sounds have line (shown in Figure 6) allows one to draw the
always been able to represent time-varying signal control functions.
flow architectures through the use of the Time- The Timeline was not so much a change to
Offset and SetDuration modules. These represent Kyma as it was the addition of an alternate view on
changes to the structure of the signal flow itself, the original underlying Sound structure. (And, judg-
not "events" in the sense of parameter updates. I ing from the number of complex, multi-layered
noticed that whenever I wanted to explain the ef- sounds and performances that have been created
fects of the TimeOffset and SetDuration in a Sound using the Timeline, it must have been a closer
(see Figure 5a), I would illustrate it using bars and match to the way people conceive of time varying
vertical time markers (see Figure 5b). structures).
Similarly, whenever I wanted to create a time-
varying synthesis architecture, I would always start Evaluating the Appropriateness
by sketching it out as a timeline on paper first. It of a Representation
was another case of finally realizing that the best At the risk of belaboring the point, all of these ex-
representation for specifying and manipulating a amples illustrate that there can be multiple ways of
structure was the one that I, along with everyone viewing and manipulating a single abstract struc-
else, had already been using on paper. ture. Each alternative can reveal or emphasize dif-

76 Computer Music Journal

This content downloaded from

132.248.9.8 on Tue, 28 Mar 2023 15:55:33 UTC
All use subject to https://about.jstor.org/terms
Figure 5a. Using Time- Figure 5b. Describing the
Offsets (Dl and D2) to cre- results of a time-varying
ate a time varying Sound Sound structure on paper
structure (Scaletti 1992). (Scaletti 1992).

I I I I I
I II I I

Mixer

I I2 S53
I I I I I

0 1 2 3 4

S1Si
delay I D2= 2 from event sources like MIDI controllers and
= 1 delay MIDI-generating software. So in Kyma version 4.5,
we added an event language, including internal and
external sources of asynchronous events and real-
time expression evaluation, on the Capybara. The
improvement in the quality of the sounds people
made in the new version demonstrated the power
of event-driven live interaction. And we were
S2 S3 pleasantly surprised that the original data structure
had been robust enough to accommodate this kind
of major shift.

ferent aspects of the structure and influence the

Expanding
kinds of things people choose tothecreate
Algorithm Set
using that
structure.

Sometimes symbols are the clearest and One mostof the ideas behind the Kyma framework (and
direct representation, and sometimes a graphical indeed behind any "modular" software or hard-
representation can be clearer and more compact. ware) is that one can easily plug new sound synthe-
Kyma provides both graphical and symbolicsis and processing algorithms into the existing
inter-
face elements, depending on the context. Arrivingstructure and immediately use them in conjunction
at the most appropriate representation nearly withal-
existing ones. Kyma has proven itself open
ways requires some experimentation and a few enough
in- to accommodate several new synthesis and
spired realizations. In the Kyma language, that processing algorithms that we have developed over
evolution is still in progress and is part of thethe years.
fun The most recent example is the addition
of participating in a living language. in 2001 of a new family of synthesis and processing
algorithms called Aggregate Synthesis.
Aggregate Synthesis is an extension to the classic
Extending the Data Structure "additive synthesis" algorithm in which complex
timbres are created by adding together the outputs
Sound objects have also proven to be extensible in of sine wave oscillators. Aggregate
of hundreds
ways I had not originally anticipated. In theSynthesis
first extends this idea by using elements other
version of Kyma, for example, although the than oscillators as the basic generators. Besides the
struc-
tures could be time-varying, the parameterclassic values"oscillator bank," one can also use a bank
were all constants. As we and other users worked of band-pass filters, a bank of impulse response
with the language, we came to realize that by mak-
generators, or a bank of grain cloud generators.
ing the parameters event-driven, we would not These alternative generator banks are controlled by
only be making the language more interactive, wethe same analysis file (or live spectral analysis) as
would also be opening it up to external control the classic oscillator bank, so Aggregate Synthesis

Scaletti 77

This content downloaded from

132.248.9.8 on Tue, 28 Mar 2023 15:55:33 UTC
All use subject to https://about.jstor.org/terms
Figure 6. Screenshot of
Kyma.5 user interface
(Scaletti 2000).

? i : i. . . . . "- I f i i .i l '" : i :i.... :..:... . . . ... . ... .. . :. . . : . " " . . ; . .:. ." , " ' . ? . ... . . . .. : N. . .. . : : " . .. . .: . . ..... . : ".. . . ... .. :.: : . ... ......

F
..
...........?r
i
.
.
... .. . - . ..C.. --,:, .... :, ..
.
.
'..::i.. " ........ .
. ..-
% :....:.:5:...:...::..".....:..::.....
.. . 11k .:1 :- . ......... .
....

....;:i:: .. . . :. .. . - :,N ... . . . . ..?::iiiiii

E;?2 `:?:.?::.:?: ,.i"O..:,:-:?::'-., :, ..: .:.. ....- .. .: . . y u . . i :'i . . . ': "
.:1..........
11.t I ...
....... .. 1::10K. ... . .. .........:.".:
.:."::.:::-.
.. . : ..i .
'' : l." .. ,: ... . .W . . " . . .. . .. . ..
. . . -I, - ::,::...:...
? ; ? - .;

. . - - -4.:..:: ?:.,:. ::.c....... .. .., . . . : ..: . .:.. - ...:.. . . . ...:.,-..::...-. .-.. .........!. ..'.. ...... . A .:!.:...:X:, -:. - :.........:: .:..... :.... %.. .
. ::

;4:--'',7 ... ......... .. .... ;.::?X .... . :::--.:--..1l..-:: -........'.: . : ,:Z.. 7::.:,?:.,? ... . ? .
.... ... .. .. - . . . . . .. . . . : . . -,::: ii: : : ::R !i X -: :-j::-l:: -.::-1: : :: , : :! :

. .., , : ,,, :?........7 ...: .:.......--........

. ...:.,::: -X o X :1-N : ??- ? .:. l .. .. . . .: : .: : . : .:

!.?..
ii . . . ?. . . .. . :.. .: : . . ..
. . . . .. . , I.,
lNl : ::::X .::: .?
::::;::' : I .."... .... I'll..... . .. . .... ... . . :.:i
.:X...."....., .. ............ ,.,.:..
,, ,,,---,:
.. . : : .. .. ; - .::--;XmX::::X:::X ... ; ... : . .
.......N? :::-::1,`:%::-.-. :!:
.i I 1. , N.""...
.. . . . . .: :l::..
. .: :...X::-..:
. ... ... : ..X...
: ......
- . ......
. ,.? .4S
.. ... ......
. . ...:: .. . . . . . .. ..".. :ii:. -'::: " ".
... 5m::%: . ..: A K ......
... . , . . .. . .:.: :: .. . ..:-..:. : :. -- . :. : ' 1BPid?: .... B .."ON&O",~~il ?

:: ,
" . , ,: . . ... ."M
. .W**000..
..... I.
% . . .. . . .. .?:?
.. . W ...
. - -...
- . ..
, . ...
. .. .:
. .. .. .
. -. .. . ! . . . . . .. : :
.. . .. - -. : : .. . . n ...:.. :.::... - :. .N.... ?
. .... ...... . ........'.4 ...
.: .. ........
.. : : - ..:,- -.::.:.: - - . .. .. .... :::. .. : ..: ..... . . :... .7. . ..: .. .... :. ...U :: N .. -- .... x: I
..- .. ... ... .. ..:: : . .:.::...
...i?..n.... .: - .-
. .:- .:.::::,
;I
:
00*! ?:--O-j~ .i6" : ...
,. . . .. .:"
.. .... :........:.. . ......:
:" ..:: .:: ? .::.:. :. b g* -bo: ' ..-
"

iii ..
.. ; ::.? ?: :% . .. .. -..... ...
:.. ,....4.....:: :
. .: :- ..n~ ~?? ?? Ilg

-:,::. ; a. :. .:: .::. .?:? : ,' " .: - ' ., : I',E?,?a:;i-i;B..,.4,kS?".,--?... ?,.- X.. :
... . . . .. ...... .. ...
......
. . ... ..D
... .........::
Irr .::: ..... . .,.. .: - - ... X.N.

, , : . . . . . . . : .:: - . . I ?.,-'.- -
: I I.I. 1..
... " -_'. - - . .. . .. - . . ?..: .:. .N. ..:.: :.::.::::::l??i .jib . . .. . .. . . ... j. : . . .. ...
., ....... . .. ....... ..-..."I.... . ::---.--:.-.;:-. I, ..... .. .... m.... :..;.........
------ Ep -- " ------_ _ ....
-- I 111I I...... iii????????????????????????????????
- ."...-..----l.,,--?.??.w?..,-.?4.;:.x?::,?-,-,,..?,.,..?.?:.:..?...?x.?:.-:?:44:..!:-".:.?L::%L?..?:.:.-.:...:Z:-?.z
.. .... ..

,_ - __ .'U .... % .: e.:.:. -X% , - , -::- .1:

!iijiii~ii~lll?BiE'"i 0?':.l; ~ ll ''" i: " i?; l~iiii !!i;-::i:i~i:.-'.i: i} i-iii.,..: .:: "?". " :?!it.'i!ii!i:"-i. .. .:ii.iii['.i~:i ii!

can be used fortiprocessor comp

live analysis/re
have a scalable
tional control parameters likear
g
envelope, graincarnation,
waveform, can su
etc.
allel processors (
24-bit 100 kHz
The Role of thespecial-purpose
Capybara
oscillator and filter circuits hardwired into a cen-
Every computer tral mixer. Its function is completely
music determined
langua
cause every computer music
by software. In our "vision-centric" society, people l
routinely dedicate
computer). In its current hardware towards the accelera-
imple
tion and improved qualitymanner
runs in a distributed of real-time video and o
graphics. Audio artists a
purpose computers: deserve the same considera-
Macinto
Capybara. The tionCapybara
(and dedicated computing cycles!).is a g

78 Computer Music Journal

This content downloaded from

132.248.9.8 on Tue, 28 Mar 2023 15:55:33 UTC
All use subject to https://about.jstor.org/terms
Language as Nexus tion of sound, visuals, tactile feedback, physical
motion, and direct neural stimulation. In these lan-
Kyma serves as a nexus connecting sound design-
guages, it will become even clearer that the map-
ers, musicians, artists, programmers, and research-
ping is the message: given identical model worlds,
ers who work in different geographical locations
the art lies in selecting or directing the observer's
and who bring with them different sets of goals and
path through that world and in mapping the num-
different ways of thinking about sound and
bers into perceptual stimuli (or electrical signals
problem-solving. They use Kyma to create sounds
suitable for directly stimulating neurons).
for feature films, advertisements, and computer
On the hardware side, I see an ever more urgent
games; they are using Kyma on albums, in stage
need for multiple, broad-band interfaces-wide
shows, for live theatre, in installations, in clubs, in
channels for pumping data into and out of the vir-
classrooms, at parties, in laboratories, in ware-
tual spaces to create a fully "immersive" experi-
houses, and at home. ence.
I learn a great deal by talking with Kyma users,
and, by inference, from watching them work,
studying the sounds they create, and analyzing the New Humans
questions they ask or new features they request.
That knowledge becomes incorporated into Kyma Hybrid Vigor
and subsequently transferred, via Kyma, to all the
other Kyma users who then further transform that Populations are shifting and, in accordance with
knowledge according to their own approaches and the principle of "hybrid vigor," all this stirring u
ways of thinking about sound. Thus, Kyma is serv- of nationalities and cultures and genes is resulting
ing as a conduit for information and knowledge in stronger, smarter, and healthier human beings
transfer between disciplines, resulting in some ex- Future computer music languages written by thes
citing cross-pollinated hybrid art forms and new ap- future humans will embody new ways of thinkin
proaches to processing and synthesizing sound. about time, space, quantity, physical movement,
music, and sound.

Future Directions for Computer Music Software Self-Modification

We are living through the early part of a century
What form will computer music languages take in when life will redefine itself through genetic eng
a future world populated by prosthetically and ge- neering and implant technology. There is no ques
netically enhanced humans having a broader con- tion that the agricultural and industrial revolutio
cept of art and an economy based more on services had a big impact on human development, but th
than on physical objects? genetic revolution is more than that. It is a redef
I foresee computer music languages that are per- nition of what it means to be human, of what it
fectly scaleable, modular, dynamic networks of dis- means to be alive. We are witnesses to the moment
tributed computational elements. They will be when the DNA molecule is just figuring out how it
customized aggregations of independent computa- can modify itself.
tional elements (some of which will be virally de- Body modifications that are now practiced as a
livered directly into our brains) that can operate in matter of style are just rehearsals for future self-
isolation or in cooperation with additional ele- modifications that will include intelligent implants
ments accessed, only as needed, via the Internet. controlled by the same neural circuits as "natural"
I see computer music languages evolving toward biological parts. The distinctions we now draw be-
more general "virtual space" languages in which a tween ourselves and our machines will be blurred
single database or model world drives the genera- and ultimately erased.

Scaletti 79

This content downloaded from

132.248.9.8 on Tue, 28 Mar 2023 15:55:33 UTC
All use subject to https://about.jstor.org/terms
New Art Under the old model, a musician or programmer
was expected to produce a physical object (a com-
Traditional distinctions between music, film, ani- pact disc) that could be distributed via trucks and
mation, and games will also become blurred. Art airplanes
of to local storage centers (stores) where
the future will consist of a multi-dimensional space consumers could purchase them. Because consum-
(literal or abstract) and a set of (deterministic orers exchanged cash for the physical object and
non-deterministic) "paths" through that space. owned the storage medium, they were led to be-
This model is equally serviceable in describing lieve a that they also owned the music or software
piece of tape music, a computer-generated anima-and could therefore freely give it away or reuse any
tion, a live-action film, a theme-park ride, a liveportion of it. Improvements in digital recording
improvisation with computer partners, a painting, a
quality and network speeds have begun to make it
novel, a computer game, a traditional piano sonata, more obvious that software and music are not
a virtual environment, a live DJ mix, or an installa- physical objects that can be purchased and owned.
tion. "Audiences" of the future will consider all of Music creation, like software creation, is a pro-
these experiences to be art works and will find the cess-an evolutionary, ongoing, iterative project.
old distinctions as foreign as the idea of sitting qui- As such, it is ill-suited to the object distribution
etly in their seats during a performance. model that worked so well for refrigerators and ba-
nanas, for example, in the last century.
The Space
The space of an artwork can be anything from a Renting the World
physical, three-dimensional space (where the What if the market for music and software were
dimensions would be time-stamped x, y, and z based on subscriptions rather than on one-time
coordinates) to a completely abstract higher- purchases? For example, one would subscribe to a
dimensional space (where the dimensions might be software provider, a favorite composer or author, or
the allowed pitch classes or timbral characteris- to a consortium of like-minded artists or
tics).
programmers. In exchange for a yearly subscription
fee, one would receive unlimited access to the lat-
The Paths est software updates, the latest pieces by one's fa-
vorite composer, or even the latest
A path through the space can be a predetermined
work-in-progress by one's favorite artist. In the case
or recorded path chosen by the artist (as in an
of an artists' or a source-code cooperative, one
acousmatic tape piece, a film, or a theme-park ride
might even be able to earn rebates on the subscrip-
where the participants sit in cars pulled along
tracks). Alternatively, it can be a "live" path (or tion fee by contributing artwork or code to a proj-
paths) where the "audience" and/or improvising ect. (This is a variation on the open-source idea in
performers explore the space interactively. which the most creative and productive members
can contribute code and the vast majority of people
who are more interested in using the product than
New Economics in creating it can contribute money.)
For one thing, this would be an acknowledge-
Software and music have much in common: both ment that in both music and software creation, the
are complex abstract structures manifested as process is more valuable than a single snapshot or
ephemeral modulations of an invisible medium.progress-report (now known as a "release"). It is al-
And unlike many commodities, one cannot buyready the case that when one buys a piece of soft-
music or software; one can license it for personal ware, one does so with the implicit understanding
use, but the ownership remains with the author. that one will be frequently downloading or pur-

80 Computer Music Journal

This content downloaded from

132.248.9.8 on Tue, 28 Mar 2023 15:55:33 UTC
All use subject to https://about.jstor.org/terms
chasing updates to that software. It would be inter- access to recorded music, "tele-performances," and
esting to extend this concept to music as well. whatever "extras" the artists choose to provide.
Instead of buying a single recording of a finished (The corollary is that the subscribers could also fire
composition, one could subscribe to the composer the composer/programmers who were not satisfy-
and access updates to that composition, alternative ing their appetites for music and software.)
performances of the same work, additional sec- If digital content were available to subscribers
tions, remixes, and new variations. The best sub- wherever and whenever they wanted, then it might
scriptions would also include opportunities to no longer be worth the effort to "pirate" or even to
interact with the creator in online master classes, store music and software. If subscribers could ac-
interviews, or group discussions. cess the content just as quickly as reading a local
For the artists and programmers, subscription hard disk, and if the content might actually im-
fees would provide a steadier source of income to prove each time they access it because the musi-
replace the current "feast-or-famine" bursts of cash cians or programmers continue to work on it, then
flow provided by the one-time purchase model. few people would find it desirable to steal the con-
Subscribers would be like "investors" in a particu- tent.
lar artist. As such, they would find it in their own In game theory, optimizing benefits in a one-off
best interest to help promote that artist to a wider interaction tends to involve maximizing one's own
audience and protect the artist against piracy gain at the expense of the other person. However,
(which, under this new model, would hardly be maintaining a longer-term relationship consisting
worth the effort in any case). of multiple transactions over some period of time
For the subscribers, it would mean better cus- requires some evidence of mutual benefit if both
tomer support, simply because an unhappy sub- parties are to continue to cooperate. For creators
scriber is unlikely to renew the subscription next and subscribers both, this new model would favor
year. Software developers already interact closely long-term, mutually beneficial relationships.
with their users and benefit from immediate feed- Even hardware could be sold by subscription
back. Would the subscription model foster more in- rather than purchased outright. This would be an
telligent dialogs on music as well? It could turn out acknowledgment that there is more value in the
that musically literate subscribers might be able to design of a piece of hardware than in the actual ma-
offer composers some genuinely useful observa- terials used. It is also an acknowledgment of the re-
tions and stimulating responses to their work (in ality that computer hardware is inevitably
addition to applause). upgraded every few years. What if, instead of buy-
The subscription model makes it more obvious ing a computer, you subscribed to a computer hard-
that the creator is even more valuable than that ware service? Every few years, the company would
which is created. Under the old model, musicians upgrade your computer to the newer model, giving
would effectively give themselves away for free on the hardware companies a strong incentive to reuse
concert tours, in interviews, bootleg recordings, existing materials as much as possible. This could
and on the web in the hopes of selling physical re- provide a steadier flow of cash to hardware compa-
cordings (that were owned by a record label, not nies while at the same time making them more re-
the musician). That made it seem as if the compact sponsive to their customers, because their
disc were more valuable than the people who cre- customers will be subscribing to a continuing ser-
ated its content (and who hold the potential for pro- vice, not making a onetime purchase. It also makes
ducing even more content in the future). Under the explicit something that is not always so obvious in
subscription model, a group of subscribers could in the current economy: the cost of reusing or recy-
effect hire the musicians/programmers, investing cling a product at the end of its useful life should
in their future output, collectively paying the mu- be factored into the initial (or in this case, ongoing)
sicians's salary in exchange for wireless network cost of the product.

Scaletti 81

This content downloaded from

132.248.9.8 on Tue, 28 Mar 2023 15:55:33 UTC
All use subject to https://about.jstor.org/terms
 It might take some small adjustments in think- all of you who are now using Kyma, thank you for
ing, but in fact we are already familiar with this collaborating with us on this lifelong project!
subscriber model in other aspects of daily com-
merce. It is the way we pay for news services like
magazines and newspapers; it is the way we pay for References
communications services like telephone and Inter-
net access; it is the way people pay for some forms Scaletti, C. 1987. "Kyma: An Object-Oriented Language
of entertainment on cable television channels. In for Music Composition." Proceedings of the 1987 In-
some sense, even paying taxes is subscribing to the ternational Computer Music Conference. San Fran-
services of the local and national government. cisco: Computer Music Association, pp. 49-56.
Scaletti,
(However, if one becomes dissatisfied with the ser- C. 1989. "Composing Sound Objects in Kyma."
Perspectives of New Music (27)1:42-69.
vice, canceling that particular subscription could be
Scaletti, C. 1992. "Polymorphic Transformations in
a bit tricky.)
Kyma." Proceedings of the 1992 International Com-
puter Music Conference. San Francisco: International
Computer Music Association, pp. 249-252.
Acknowledgments Scaletti, C. 1997. Kyma Sound Design Environment.
Champaign, Illinois: Symbolic Sound Corporation.
Special thanks to Kurt J. Hebel, my longtime Scaletti, C. 2000. Kyma.5 Walkthrough: A Tutorial In-
collaborator without whom Kyma and Symbolic troduction to Kyma.5. Champaign, Illinois: Symbolic
Sound would not exist in their present form. Sound Corporation.
Thanks to Jean, Paul, Matt, Barry, Perry, Robin, Scaletti, C., and K. J. Hebel. 1991. "An Object-Based Rep-
Mike, Donald, Desiree, Chip, and Mark who have resentation for Digital Audio Signals." In G. De Poli et
al., eds. Representations of Musical Signals. Cam-
helped run the office, ship systems, and build hard-
bridge, Massachusetts: MIT Press, pp. 371-389.
ware at Symbolic Sound. Thanks to the charter Scaletti, C., and R. E. Johnson. 1988. "An Interactive
members of the Kyma community: Dick Robinson, Graphic Environment for Object-Oriented Music Com-
Francesco Guerra, Richard Festinger, Brian Belet, position and Sound Synthesis." Proceedings of the
John Paul Jones, Alan Craig, Frank Tveor Norden- 1988 Conference on Object-Oriented Programming
sten, Lippold Haken, Stu Smith, David Worrall, Sal Languages and Systems. New York: ACM Press,
Martirano, Joran Rudi, and Bruno Liberda. And, to pp. 18-26.

82 Computer Music Journal

This content downloaded from

132.248.9.8 on Tue, 28 Mar 2023 15:55:33 UTC
All use subject to https://about.jstor.org/terms