ï~~Proceedings of the International Computer Music Conference 2011, University of Huddersfield, UK, 31 July- 5 August 2011
STRING, A DIRECT AUDIOVISUAL MODEL
Benjamin Schroeder
The Ohio State University
Dept. of Comp. Sci. and ng.
benschroeder@acm.org
Marc Ainger
The Ohio State University
School of Music
ainger. l@osu.edu
Richard Parent
The Ohio State University
Dept. of Comp. Sci. and Eng.
parent@cse. ohio-state. edu
Physically based sound models can generate complex, expressive sounds, but have correspondingly complex definitions. Interactive visual representations can help explain and reduce this complexity, showing the relation between a model's definition and its behavior. Furthermore,
such representations allow for straightforward connection
of different models. This kind of interactivity is present
in many computer music languages, but its applicability
is limited in the case of physical models.
We present the Visible String, a real-time audiovisual
model of a physically based string. Users can see and hear
the output of the string model and can use simple analysis
tools to investigate further. The model's parameters may
be viewed and changed while the string is being played.
Gestural input both plays the string and drives the parameters of an associated plucking model. Using these tools,
users can build intuition about how the models work. Finally, individual strings may be connected to one another
and to external sources to form modular audiovisual synthesis networks, extending both synthesis capabilities and
analytical interest.
1. INTRODUCTION
Computer music languages, especially visual ones like
MaxMSP 161 and PD [71, have long encouraged interactive exploration. One of the greatest strengths of these
languages is the ease with which a programmer can create
and connect synthesis objects, change their parameters,
and inspect (and hear) partial results by inserting debugging objects, all without stopping the running program.
These facilities help one to quickly gain understanding of
new objects and unfamiliar programs.
Physically based sound models have great promise:
they can create expressive, realistic sounds and have parameters that are rooted in familiar real-world concepts.
However, it is difficult to use the full power of these models, or to understand their mathematical foundations, using traditional computer music languages.
Physical models may be used as "black box" unit generators in traditional languages by choosing some set of
attributes and inputs and producing a set output. For example, a guitar string unit generator might support the selection of material properties, frequency, a plucking point,
and a pickup point, and produce a continuous sound signal
taken from the pickup.
However, this limits the way in which such a model
can be used-in the example above, it would be difficult
to change the excitation model without changing the internals of the object, or to pick up sound from more than one
point. It also separates the model from its mathematical
basis as a continuous object in a physical space. Computer languages meant for engineering (such as Matlab)
have the opposite problem: they have a direct connection
to the mathematical model, and are flexible, but they do
not easily support visual interaction and in any case are
not built around musical concepts.
In this paper we present a way to make interaction and
analysis of physical models concrete in the way that interaction with unit generators in visual computer music
languages is concrete. The Visible String provides direct
audiovisual interaction with a musical string and its mathematical representation.
UAL
y,,,......,.
4 2 -tp h..ri Is.- -
i...ure 1. A sing and its equatiot ot motiot
Figure 1.A string and its equation of motion.
328
0
