A psychomechanical study of rolling sounds
Résumé
In this thesis both the mechanics and the perception of the sounds from rolling and bouncing objects are studied. It is necessary to combine these two research domains, resulting in what is called psychomechanics, because only in this combination can we discover the acoustical origin of the information a listener uses to detect mechanical properties of objects when listening to the sounds generated by these objects. In this thesis objects are balls rolling over or bouncing on a plate. The characteristics of such interactions are de ned by physical properties of the components in the interaction, among them the size and the speed of the rolling balls, and, perhaps, the thickness of the plate over which the ball rolls. The research question is to what extent the listener can "hear" these physical properties and on what kind of information in the sound this is based. The rst two chapters of this thesis investigate bouncing while the remaining ve chapters deal with rolling sounds.A typical parameter in investigating bouncing processes is the restitution coef cient, de ned as the proportion of two subsequent time intervals between bounces. The traditional model for calculating the restitution coef cient of a ball was adopted by calculating the plate response from the point impedance of the plate. This analytical model was compared with a numerical model and with experimental results. The restitution coef cient is one source of information the listener may use to detect the size of the bouncing ball; the spectral content of each impact sound is another. The in uence and relative weight of each of these two sources of information was investigated in a perception experiment. The numerical model for simulating a single impact of a ball on a plate was adapted to simulate the movement and the sound of a ball rolling over a plate. The main changes were made in the contact between the ball and the plate. Instead of an impact point that was xed in space and short in time, the model now incorporates a contact point that is variable in space and continuous in time.Furthermore, the model was extended to contain the surface roughness of the plate. The validity of the model was demonstrated by simulating the ball and plate movement, and comparing the results with experiments and analytical calculations. In the case of a ball moving over a plate, a temporal effect was observed due to the interference between the sound directly generated at the point of contact between ball and plate, and the sound re ected at the edge of the plate. This effect was added to synthesized rolling sounds which resulted in a more natural sound. In a related perception experiment it was shown that the acoustic differences in the sounds of balls rolling towards or away from the edge of a plate are suf cient to perceptually discriminate these sounds. Listeners were, however, not able to indicate the rolling direction of the ball. That is, listeners could not interpret the acoustic information correctly in terms of perceived mechanical features. The auditory capabilities to perceive the size and speed of a rolling ball were studied in a difference-scaling task, where the participants were asked to judge the difference between two sounds without having to label the origin of these differences. In additional experiments participants were asked to rate explicitly the mechanical properties of ball and plate. The experimental methodologies used were paired comparison and absolute magnitude estimation. From these results the distances in the perceptual space of rolling sounds were derived. Furthermore, by calculating the psychometric function for the participants' estimations, an exact measure for the probability that a listener correctly identi es the difference in one of the physical parameters was calculated.The in uence of some object properties, here notably the plate thickness, can be detected in a difference scaling task, but listeners can not estimate the thickness in a magnitude scaling task. This indicates that the auditory perception of the mechanical object properties is a layered process, and that detection and estimation nd place at a different level of such a layered process. On the other hand, for the two remaining mechanical properties, the size and the speed of the ball, the estimation of the size was more robust, in the sense that is was little in- uenced by a change of experimental methodology or simultaneous variation of the other mechanical parameters.
Pas de réumé disponible en français
Loading...