~Proceedings ICMCISMCI2014 14-20 September 2014, Athens, Greece
Understanding and Tuning Mass-Interaction Networks
Through Their Modal Representation
Jerome Villeneuve
ICA laboratory, ACROE,
jerome.villeneuve@imag.fr
ABSTRACT
Sound is all about vibration, and the GENESIS environment provides an efficient way for modeling and simulating complex vibrating structures, enabling to produce rich
sounds.
In this paper, we propose an overview of tools recently developed and available within the GENESIS environment, allowing a better understanding on how massinteraction networks behave and introducing some enhanced tuning of their vibrating properties. All these tools
try to address an inherent need of any creative process
either in the physical world or in GENESIS, which is to
create bidirectional connections between properties of a
phenomenon, in our case, audible sounds, and properties
of what produced it, here, mass-interaction networks.
For this purpose, we will introduce the topological
and modal representations of such mass-interaction networks and appreciate how relevant it can be to switch
between these different representations to really apprehend its inner properties and those of the sounds it produces.
1. INTRODUCTION
GENESIS [1] is a musician-oriented software environment for sound synthesis and musical composition. It
implements the modular language CORDIS-ANIMA [2]
that provides a set of primitive elements, modules satisfying Newtonian physics.
The elementary bricks of CORDIS-ANIMA are distinguished in two categories: the <MAT> modules that
designate "matter" being either moving or fixed and the
<LIA> modules that designate interactions and allow
interconnections between <MAT> modules.
A GENESIS "topological" model is then constructed by
connecting these modules to each other into a dedicated
2D space called "the bench", this by direct manipulation
(Figure 1). Such a model will be defined by its "topology", meaning its network structuration, by the parameters
of each one of the modules it is made of (M for inertia of
MAT modules, K for stiffness, Z for damping for LIA
modules), and finally by a set of initial conditions carried
by MAT modules (position and/or velocity).
Copyright: ~ 2014 J. Villeneuve and al. This is an open-access article distributed under the terms of the
which permits unrestricted use, distribution, and reproduction
in any medium, provided the original author and source are credited
Claude Cadoz
ICA laboratory, ACROE,
Claude.Cadoz@imag.fr
Once built and parameterized, the model will be set in
motion through an outer excitation, and its movement
will be ((listened to (Figure 1).
Figure 1. Mass-interaction network of a "string" (indeed, its geometry is quite unusual, but it does not matter. Its topology does). Yellow dots are moving
<MAT>, green ones stay fixed.
From there, one can build simple or very large massinteractions networks, composed with thousands of modules. Once simulated, these allow obtaining complex and
rich sounds. But whatever the size of such models, when
it comes to exploring the variety of sounds they are able
to produce, or even to adjust their nature or structure with
the aim of fine-tuning this sounds, a major problem arises. Its widest formulation expressing a need to establish
two-ways relations between causes and effects.
There are several complementary ways to go on with
such concerns.
The first, empirical, is to build one's own knowledge
of how objects, models, behave according to how we act
on or modify them. One will have to learn how slight
adjustment of model parameters will alter perceptive
properties of the product of its simulation. Furthermore,
in the case of the GENESIS environment, it is now possible to physically experiment virtual models by means of
haptic interfaces [3], which allow, by an additional gestural feedback, a very direct and instrumental way to
investigate the properties and possibilities of models in
terms of musical creation. In both cases, a closed loop is
- 1490 -
0