Information Technology Reference
In-Depth Information
enabling an expert or an intuitive sound manipulation. Finally, as a perspective
to this survey, explicit and implicit brain-computer interfaces (BCI) are
described with respect to the previous theoretical frameworks, and a semiotic-
based BCI aiming at increasing the intuitiveness of synthesis control processes is
envisaged. These interfaces may open for new applications adapted to either
handicapped or healthy subjects.
4.1
Introduction
In this article, we present different approaches to sound synthesis and control issues
and describe how these procedures can be conceptualized and related to different
paradigms within the domain of cognitive neuroscience. A special emphasis is put
on the notion of intuitive control and how such a control can be defined from the
identi
cation of signal invariants obtained both from the considerations of the
physical or signal behaviour of the sound-generating sources and the perceptual
impact of the sounds on the listeners.
Since the
first sounds were produced by a computer in the late 1950s, computer-
based (or synthesized) sounds have become subject to an increasing attention for
everyday use. In early years of sound synthesis, the majority of applications were
dedicated to musicians who learned to play new instruments that generally offered a
lot of control possibilities, but required high skills to operate. Due to increasingly
powerful computers, new applications linked to communication, virtual reality and
sound design have made sound synthesis available for a broader community. This
means that synthesis tools need to be adapted to non-expert users and should offer
intuitive control interfaces that do not require speci
c training. The construction of
such intuitive synthesis tools requires knowledge about human perception and
cognition in general and how a person attributes sense to sounds. Why are we for
instance able to recognize the material of falling objects simply from the sounds
they produce, or why do we easily accept the ersatz of horse hooves made by the
noise produced when somebody is knocking coconuts together? Is the recognition
of sound events linked to the presence of speci
c acoustic morphologies that can be
identi
ed by signal analysis? In the approach presented here, we hypothesize that
this is the case and that perception emerges from such invariant sound structures,
so-called
invariant sound morphologies
, in line with the ecological approach of
visual perception introduced by (Gibson
1986
). From a synthesis point of view, this
theoretical framework is of great interest, since if enables the conception of per-
ceptually optimized synthesis strategies with intuitive control parameters.
Sound synthesis based on the modelling of physical sources is generally divided
in two main classes, i.e. physical models and signal models. Physical models aim at
simulating the physical behaviour of sound sources (i.e. the physical origin of
sounds), while signal models imitate the recorded signal using mathematical rep-
resentations without considering the physical phenomena behind the sound pro-
duction. In the case of physical models, an accurate synthesis can only be achieved
when physical phenomena linked to the sound production are well described by
