A model for human-computer interaction based on the recognition of musical gestures

Presents a model for human-computer interaction based on two-level automatic recognition of musical gestures. The basic idea underlying this model is that each gesture can be performed with different nuances, which carry some additional information about the intentions and feelings of the performer. The communication can then be achieved through the recognition both of the class to which a gesture belongs and of the nuances applied to the gesture. The model is applied to the recognition of gestures on a musical acoustic instrument. As a test instrument, the classical guitar was chosen. In the proposed model, the interaction is achieved without the use of sensors but through the analysis of the characteristics of the audio signal produced by the instrument. Hence, by simply analysing the sounds produced by performer's gestures, the model is able to recognize both the kind of gesture a guitarist is performing (i.e. the note played with given pitch and loudness) and the nuances of this gesture (i.e. the way the string is plucked)