Searching for Cross-Individual Relationships between Sound and Movement Features using an SVM Classi er

Abstract

In this paper we present a method for studying relationships between features of sound and features of movement. The method has been tested by carrying out an experiment with people moving an object in space along with short sounds. 3D position data of the object was recorded and several features were calculated from each of the recordings. These features were provided as input to a classifier which was able to classify the recorded actions satisfactorily, particularly when taking into account that the only link between the actions performed by the different subjects was the sound they heard while making the action.