Information Technology Reference
In-Depth Information
physics. This is not the case for complex sources (e.g. natural phenomena such as
wind, rain,
fire, etc.). In the case of signal models, any sound can generally be
perfectly resynthesized for instance from the analysis of real sounds, independently
of the complexity of the underlying physical phenomena of the sound source.
However, the control of such sounds is a dif
cult issue due to the large number of
synthesis parameters that generally are implied in such models and to the impos-
sibility to physically interpret these parameters. The physical and signal models can
also be combined to form so-called hybrid models (e.g. Ystad and Voinier
2001
).
The control of these models requires an expertise and the quality judgment of the
control is based on an error function linked to the physical or signal precision
between the model and the real vibration. Such controls necessitate a scienti
c
expertise apart from certain cases such as musical applications where the control
parameters correspond to physical values controlled by the musician (e.g. pressure,
force, frequency, etc.). In this latter case, the musical expertise enables the control.
To propose efficient synthesis models that enable intuitive control possibilities,
synthesis models combined with perceptual considerations have been developed
lately. Perceptual correlates have been sought by testing the perceptual relevance of
physical and/or signal parameters through listening tests (cf. Sect.
4.3.2
). In the case
of environmental sounds, we have identi
ed such perceptually relevant sound
morphologies through several experiments. These experiments have made it pos-
sible to identify sound elements, also described as sound
“
atoms
”
, speci
c to given
sound categories that enable de
nition of high-level control parameters for real-time
synthesis applications. Such synthesis tools allow users to synthesize auditory
scenes using intuitive rather than re
ective processes. Intuitive processes appeal on
intuition which is a kind of immediate knowledge, which does not require rea-
soning, or re
ned as the knowledge of an
evident truth, a direct and immediate seeing of a thought object (Lalande
1926
). The
quality of the control strategy is in this case based on perceptual judgments and on
easily understandable control parameters on the user interface. Therefore, we call
this synthesis control,
intuitive control
.
When searching for perceptually relevant sound morphologies, the understand-
ing of attribution of sense of sounds becomes essential. This issue is a natural part
of a more general research
ective thought. Intuition can also be de
field called semiotics that consists in studying the
general theory of signs. The notion of signs has been addressed since antiquity by
the stoic philosophers (Nadeau
1999
). Classically, semiotics is divided in syntax,
semantics and pragmatics. Semiology is a part of semiotics, which concerns the
social life, and dynamic impact of signs, as language (Nadeau
1999
). For
de Saussure, language constitutes a special system among all semiological facts. In
linguistics, for de Saussure, a sign is the association of a signi
er (acoustic image)
and a signi
ed (the correlated concept) linked together in a consubstantial way
(de Saussure
1955
). This consubstantial relationship is often dif
cult to understand.
Semiotics span over both linguistic and non-linguistic domains such as music,
vision, biology, etc. This means that it is possible to propose a semiotic approach of
sounds, without referring to linguistic semiology. Like in de Saussure construction
of signs, one can postulate that every natural (environmental) or social sound is
