Timbral Composition and Gesture
Timbral Composition and Gesture
Columbia University Music Department
709 Dodge Hall, Columbia University, New York, NY 10028
ABSTRACT: A stochastic granular synthesis instrument shaped
the use of timbre in the composition, "Dragon of the Nebula".
The changing grain parameters within each event create timbral
gestures, which make up the musical phrases. These gestures
may be represented in a two-dimensional graph. The perception
of these gestures and the relationship to the performance situation
will be discussed.
Dragon of the Nebula, for computer generated tape, was written after composing
several pieces with timbre as an important element, one using often pitched
frequency modulation sounds and one using dense sampled timbres. In this
piece, however, the sounds are all synthesized, precisely controllable, and
primarily unpitched. A granular synthesis instrument was used to produce events
called gestures, which interact in various ways. The piece is structured by the
timbral relationships which evolve throughout the piece.
Since I first started composing electronic music, I had wanted to do a piece that
was simply a transformation of "one sound", with all the flexibility of timbre and
other parameters available with analog synthesizers. This "one sound"
connectedness and continuity through extreme timbral diversity could be a
musical analogy to some of the philosophical ideas of the interconnectedness of
beings, and the manifestation of self in different forms from Taoist and Buddhist
ideas that I had explored. Whether or not this idea would work musically, it
seemed to be an interesting sound experiment. It was difficult to get the control
over parameters which would allow subtle and gradual change while still
maintaining the singularity or continuity of the gesture. In Dragon of the Nebula
the goals were somewhat different, but I still wanted this timbral diversity with
continuity, particularly in the final section.
A granular synthesis instrument, "sgran", made this control more realizable. For
a description of granular synthesis techniques see the Roads (1988) and Truax
(1988) articles. The approach is derived from Xenakis' music in that stochastic
arrangements of grains are used, but it is not as mathematically-driven as some
of his music, since I am more concerned with the individual sound shapes
desired or created rather than with the field of possibilities produced under
certain mathematically described conditions. His graphs of lines indicating new
form possibilities in Formalized Music suggest some of the ways these musical
shapes can be visualized. "Sgran" is an instrument which allows precise timing
control of the grain characteristics, written in Paul Lansky's Cmix programming
language. A NeXT graphical interface, StochGran makes the instrument, with its
thirty-seven parameter fields per event, easier to use. For a more complete
description of these programs, see Helmuth (1991).
The "sgran" instrument user composes a bunch of curves which define each
parameter, all of which are simultaneously and independently unfolding
throughout each event. These "events" can be layered or serial, overlapped or
contiguous, with the Cmix "mix" program. This process gives one a lot of
freedom, particularly since the possible sounds cover such a wide range. It is
possible to visual a frequency / space correlation and represent the changing
frequency bands as high or low in space, and narrow or wide according to the
frequency band. The variation can be thought of in terms of a "fuzziness" of the
lines. Grain regularity, rate, envelope and duration have less clear
correspondences to a visual diagram, but one can indicate a "roughness" or
"smoothness" in texture which might indicate holes between short grains or
smoothly overlapping long grains. The gesture is mapped onto a graph as a
musical shape, a geometric form with a particular texture. It is possible to graph
a sound out in this way:
example: A high very narrow frequency band with a smooth sound slowly and
unevenly descending to a low wide rough band.
The synthesis instrument generates grains of sound stochastically with frequency
and timing controlled by the composer. The timing parameters affect what the
listener hears as density, timbre and "smoothness" of sound. The grain timing
and frequency parameters are specified for the beginning and end of an event,
each with a changeable amount of variation allowed. The event that is created
may change in any parameter according to a function specified by the composer.
When these sounds can change over time with a comparative amount of
freedom, they can take on the subtlety of a musical motive or phrase.
The musical unit that is created can be called a "gesture" or "event-shape".
While it may be heard as a "phrase", often the musical phrase may contain many
of these units. "Event-shape" indicates the visual representation that is possible
of mapping the parameter changes onto a graph. The word "gesture" has
meaning in composition particularly in music that deals with timbre and gradual
change rather than in a note-oriented approach. Of course, a particular
modulation or cadence could be described as a gesture, but it could be described
more accurately as just that modulation or cadence. A gesture is something
more than a change in one parameter. It is a combination of changes producing
a unique unit of sound. This unit may be related to a geometric form or other
visual image in the composers or listener's mind, or may be represented in color
changes or a physical movement. For example, conductors or other musicians
will often half sing, half physically gesture to communicate a part of the music
they are talking about, or an interpretation of the part that they want to hear.
Sometimes the pitches are not sung accurately, but because of the combination
of the rhythm, pitch contour, dynamic changes, or articulation it will be clear what
is indicated. It is the gesture that is described. Analysis of tonal or serial music
usually deals primarily with pitch and rhythm. In timbral music there may be too
many parameters to easily select one for analysis. The interaction of several
more subtle parameters may define the event. A sonogram can certainly be
helpful, but it contains more information than necessary. Sometimes rhythm can
be difficult to read, and harmonic pitch relationships are often lost. A composer
develops ways of categorizing and relating timbres in some intuitive or systematic
way. To be aware of the units in which a composer is thinking is half way toward
beginning to understand the music.
To create Dragon of the Nebula I generated a large number of sound files initially
as I worked on each section. To keep track of and organize them I gave
individual gestures names containing words such as "metallic", "sparse", "noisy"
to indicate something about the timbre and density, and I worked with pictures of
the frequency changes and other characteristics. I did not create a highly
organized catalog of visual maps such as the one in the above example, but I did
think of the sounds in terms of that type of representation when appropriate. The
visual mapping would sometimes suggest ideas which then could be synthesized
and evaluated. Gestures were related by means of symmetry and contrast.
The form of Dragon of the Nebula is essentially through-composed. The four
sections each have very different materials. The beginning section is a series of
noise bands, which descend in frequency irregularly ending in low explosive
sounds. The second contains large grains heard as individual microtonal pitches
with different type of room simulation, and the third is dense bands of sine waves
with a thick chorusy sound. Last is the "dragon" section, a linear, fast
succession of dramatically varying timbres grouped into phrases,* separated by
silence. The fast changes were to convey my impression of the chaotic,
ungrounded and destructive energy in the international situation during the Gulf
War. The phrases were shaped with linear changes in frequency, density, or
timbral "roughness", sometimes punctuated by sharply opposing gestures. In the
earlier sections musical shapes were constructed intentionally and these phrases
tended to be longer and more slowly changing over time. In the later sections the
phrases were generated algorithmically with a score generator. The algorithmic
approach achieved maximum variation in timbres in a short time, especially when
layers were produced and then mixed algorithmically as well. While some
algorithmic methods were used, on the larger level the gestures were connected
intuitively in all sections. The amount of timbral change steadily increased
throughout the piece. One reason that a through-composed form, with sections
and whole forms that under-go gradual transformations is natural, may be that
the smaller level units also change gradually, rather than being made up of note
patterns which repeat. The principle of continuous change is reflected both in the
individual gesture and in the overall form.
In instrumental music, one must resort to techniques that instruments were not
designed to do to get timbral variety. In concrete music one must work within the
confines of the acoustically produced sound. One can't alter physical laws to
make sounds, for example produce the sound of a water drop falling half as fast
as it would on earth. While signal processing algorithms make many things
possible, they often have side effects. One can either juxtapose sounds or layer
them, but to achieve a true continuous sound transformation from one timbre to
another may be impossible. This has been one criticism of timbral composition:
that the parameter is not controllable enough to lend itself to sophisticated
composition. With the granular synthesis instrument, however, timbral phrases
can certainly be manipulated in time, frequency range, location and other
parameters. One can move from any sound to any other sound, at any rate
desired. Also, the changes in parameters may be made at different rates
according to different functions. Precise timbral control is possible over time.
Relationships between the parameters can be composed, as other composers
have composed with timbre by working with acoustic instrument "families", or
classifying frequency modulated timbres according to prominent frequencies as
Dashow has done. If this is done with some skill this may defeat another
criticism of timbral pieces as "slow" or "boring". This negative perception can
come from a lack of listening to the more complex timbral changes, perhaps due
to musical training which does not consider timbre any farther than a course in
some details of orchestration. Much depends on the what aspect of the music
the listener is focusing on. The rate of information in a timbral change might be
great, and the shape elegant, but if the person is listening for "notes" s/he may be
In this music a connection is made between all of the aspects of the sound; the
"parts", or parameter changes, coordinate rather than operate independently.
The counterpoint is within the interaction of the parameters of the sound itself.
This one-pointedness or focus creates an intensity of experience of the sound of
the moment, when the possibility of change is so great. An example of the
experience I mean can be heard in shakuhachi playing, which has the intense
contrast between the noisy air sounds and the pure wood tones. Both timbres
shape the phrase. The sounds are different timbrally, yet both produced by the
same performer and instrument and consequently intertwined in the physical
structure of the flute and human body. In the case of computer generated music
which has no live performer, the connection between performance and
experience, or performer's movement and musical gesture is subject to the
audience's imagination. The audience can receive the impression of a "super-
instrument" or perhaps be released from the impression of any particular
mechanical instrument at all. The gestures are freer, and make use of a broader
timbral palette, with the ability to move between any of the timbres in the palette
as a painter could mix any colors together. Previously these sounds could only
be layered or juxtaposed. The intention behind this is greater expressiveness.
How does the audience relate to this music without a physical performer, or no
longer even the impression of remembered performance when a familiar
instrumental sound is physically modelled as in the Karplus-Strong plucked
string algorithm? An expanded expressivity can come from the abstract sound
created without mechanical constraints. Music is a communication of the mind,
not a sport. The gestures can express ideas independently of any physical
gesture. The composer can create his/her own associations within the
composition. This could allow more complex ideas to be expressed without the
vestigial artifact of performers' arbitrary movements. To express ideas well, the
composer must be free of constraints in the medium, and in the performance
situation. With the techniques described above, many constraints are removed.
Also, the need for the larger, perceived "grain" in the medium, the note of a
classical symphony, or the splice of analog music is gone. The need for the
layering/mixing of concrete analog or computer music is unneeded. This is not to
say that rhythm, notes, or layers can not be used successfully, but that it is no
longer necessary to construct music in those units. It is possible to explore pure
Dashow, James. 1980. "Spectra as Chords". Computer Music Journal, v. 4, p.
43 - 52.
Helmuth, Mara. 1991. "Patchmix and StochGran: Two Graphical Interfaces".
Proceedings of the International Computer Music Conference. Montreal:
Computer Music Association.
Roads, Curtis. 1988. "Introduction to Granular Synthesis". Computer Music
Journal, v. 12, no. 2. Summer, 1988. Cambridge, MA: Massachusetts Institute of
Truax, Barry. 1988. "Real-Time Granular Synthesis with a Digital Signal
Processor". Computer Music Journal, v. 12, no. 2. Summer, 1988. Cambridge,
MA: Massachusetts Institute of Technology.
Xenakis, Iannis. 1971. Formalized Music. Bloomington, IN: Indiana University