Modelling for sound creation

Spatialisation (Claude Cadoz)

Projection of sound sources in computer music using loudspeakers induces, generally speaking, a dimension what is conveniently called "spatial". The term denotes the capacity of the auditory system to associate, to some extent, sound stimuli of localisation attributes (distance, azimuth, elevation), of multitude, and of motion, to the sources producing them. Loudspeakers sound systems, referred to as "spatial", convey these capacities on the base of several principles and technics (from simple stereophony to wave fronts synthesis).Vibratory systems produced with CORDIS-ANIMA formalism display per se as many source points. In addition, the topological structure of their interactions could engender a similitude with a coherent spatial dissemination. Source points interaction principles of every loudspeakers of a system can therefore be actionated to address analytical abilities of sound sources sytems of the auditory perception. Thus, sound spatialisation shall not imitate reality, yet shall explore at most engendered models signs in this auditory perception modality, transforming them to "forms conteners" for the musical discourse. Four different technics have been explored:

  1. The easiest, employed naturally for the first simulations. This simple method assigns a specific headspeaker to every points of a simulated moving points subgroup of a vibrating structure. It results in an extension feeling of the vibrating structure in the space between the loudspeakers.
  2. The "Azimuthor", using a range of "trestles" (hyper-trestles) gathering vibrations of every vibrating structures attached and which, each using their own amount, send by visco-elastic transmission a proportion of signals to the whole loudspeaker complex. Each "trestle" could thus be considered as a source point localised precisely in azimuth.
  3. A third technique consists in transmiting vibrations of every individual source points to a specific calculated transmission coefficient with a mathematic model external to the situation, in order to handle the consistency check of the loudspeakers and the energetic consistency. The modelling process stays physical, if we may say, yet the physical rules calculations are external. Any systems of spatialisation developped can take over this approach.
  4. A fourth technique comes close to the principle of simulation in a propagation environment: by reprensenting the environment where the source is, first by simplifying it supposing it only two-dimensional, then by simulating this 2D with a masses-interactions network shaped like a two-dimensional topological slab. Thus, sources can be placed anywhere on this slab area and transducers (loudspeakers) can receive movement from any slab points, corresponding to different places in the real space.

Any of these technics have their own advantages and limits. The 2012 achieved work on this point consisted in a whole methodical experiments and highlighted the complementarity of these approaches, which can therefore be explored in a relevant way in a musical construction. This work is synthetized in technical records to be achieved in 2013, as part of GENESIS didactic environment development.

Modal configuration of wide vibrating structures (Claude Cadoz)

As soon as 2011, the increase of simulation powerness engendered by the latest GENESIS developments enables to reach a new research issue on "wide" vibrating structures modal properties, according to some of their global features: "shape", "netting" type, edges state, etc. The "modal configuration", id. for a given vibrating structure, the shape of frequencies disseminations of the structure modes has a direct cause, though difficult to analyse, on the nature of the sound. Besides, if it is possible to classify different categories of harmonic sounds, produced by lineal structures (one-dimensional topology as part of GENESIS), it is particularly complex as part of wide range of non-harmonic sounds produced by two-dimensional structures. The aim is therefore to establish some categorised spotting according to perceptive features relating to relative features of the factor structure. In 2012, specific exploring and experimentation methods were achieved, correlating modal structural properties (shape, netting, edges type, etc.), with their modal properties (the spectrum structure they produce) and the perceived qualities, with their perceptives correlations. This work is synthetized in technical records associated with didactic models and sound samples, also to be achieved in 2013, as part of GENESIS didactic environment development.

Studies of interactions between signal damping ratio and viscosity types (Olivier Tache, Audrey Pellissier, Jérôme Villeneuve)

The damping structure of vibratory modes is another important property of the vibrating structures feature, alongside with the structure frequency considered in the above studies. Generally speaking, modal properties of a linear vibrating structure can be defined and be observed from their "modal analysis" relating the frenquency, the amplitude (in specific excitation and hearing conditions) and the signal dampling of each mode. From a matrix mathematical transformation, it gives of the whole general linear masses-interactions a representation in the shape of an independant range of second range linear oscillators, quantify by their frequency and their own damping ratio, or, to put it another way, their own physical parameters of elasticity and viscosity (K and Z in GENESIS). GENESIS has a specific feature (very often used in practice), enabling on the one hand to make the modal analysis of any GENESIS network <MAT><LIA> on the condition it shall be exclusively linear, on the other hand to "tune" a network by imposing one (and only one) value in the analysis tab, whether it is constant frenquency or damping ratio stiffness or viscosity. Yet, two particular damping situations are revealed: one where damping coefficient are homogeneous (all identicals) within the model in its topological description, and another one where they are homogeneous in the modal description. This situation shows a certain interest since the first may be read as corresponding to a viscosity within the vibrating structure, the second as a central viscosity in which the vibrating structure is in. Perceptives correlations of these two situations are relatively coherent with this metaphor. Olivier Tache takled this issue through Audrey Pellissier's engineering final year training, in 2012. It enables producing methods in GENESIS using PNSL2 script language, controlling precisely such features and particularly to forecast, to some extent, an independant timber control, in its causal reference, and the persistence and resonance of a vibrating structure.

Modelling for macro-structural creation (Claude Cadoz, Giuseppe Gavazza)

CORDIS-ANIMA formalism enables to produce masses-interactions networks having dynamic properties at an infrasonic scale. From this, a merger is possible from temporal forms it may engender and the macro-temporal structure of the flux of musical sound. More generally, on the condition of introducing non-linear interactions within a CORDIS-ANIMA model (which is generally possible with the formalism, if well controlled), an energy transfer can occur from slow evolving components (sub-audio, or on a gestural mechanicals behaviors scale) towards faster behaviours components (acoustics).Thus, there are possibilities to include gestural instrumental behavior modelling within the model, and widely, to give to its components having a slow behaviour the power to produce sound flux form development in the long run, by staying into the physical representation paradigm. More generally, from any model composed of linear and non-linear components, which parameters values scaled in every range of authorised variation (assuring the reliability), can emerge macro- and micro-structual sound dynamic phenomenons. It is then to ascertain, in a direct way, structures giving relevant macro-temporal phenomenons, in the other way, structures giving predefined macro-temporal forms. C. Cadoz explored and exploited a first range of basic complementary processes (for his piece pico..TERA, 2002). In 2012 this work is taken back again, constituting the core issue of Giuseppe Gavazza's thesis at the laboratory. Started in October, the thesis is named Synthesis through physical modelling as a musical formalism tool ("La synthèse par modèle physique comme outil de formalisation musicale").