Sound synthesis as a way to access to the sense of sounds

The way we attribute sense to sounds is still a unknown process. How do we identify merely by listening that someone has knocked at the door, or what in music which evokes us various emotions ? The attribution of sense to the sound event is enabled by some specific cues contained in the acoustical signal. This study consists in better understanding the relationship between the characteristics of the acoustical signal and the mental representation associated to the perceived sound, and more precisely in specifying the acoustical cues conveying the pertinent information. For that purpose, an efficient tool is the sound synthesis. The main advantages of using sound synthesis techniques consist in manipulating precisely the signals and in constructing calibrated stimuli for pertinent experiments in psycho-acoustics and cognitive neurosciences. We presented an example of study which associates sound synthesis, behavioral and brain imaging methods. The aims were to examine the neural bases of the processes involved in the categorization of different impacted materials (Steel, Wood or Glass) by measuring the electrical brain activity (Event Related Potentials method), and to relate these results to the acoustical parameters of sounds in order to point out perceptually and cognitively pertinent parameters for an ecological synthesis of these sounds. For this experiment, natural sounds were recorded, analyzed and resynthesized, and a sound morphing process was applied to construct hybrid sounds simulating progressive transitions between different materials. Electrophysiological data revealed that the temporal dynamics of the processes associated to Steel sounds differed significantly from Glass and Wood sounds as early as 150 ms after the sound onset. Thus, Steel sounds may be differently processed by the brain compared to Wood and Glass. From an acoustical point of view, a timbre model based on the characteristics of the ERPs was proposed. In particular, an acoustical analysis showed that damping and tonal consonance were pertinent acoustical cues, as they would explain the categorization processes performed by the brain. Furthermore, the determination of typical sounds for each category allowed for the calibration of the synthesis model parameters linked to the damping. Thus, we proposed a more intuitive control of the synthesis model using semantic description of the materials. It is first step towards the problematics related to the mapping.