URL:

files/proceedings/2008/session9_number1_paper14.pdf

Abstract:

The work described in this paper is part of a project that aims to implement and assess a computer system that can control the affective content of the music output, in such a way that it may express an intended emotion. In this system, music selection and transformation are done with the help of a knowledge base with weighted mappings between continuous affective dimensions (valence and arousal) and music features (e.g., rhythm and melody) grounded on results from works of Music Psychology. The system starts by making a segmentation of MIDI music to obtain pieces that may express only one kind of emotion. Then, feature extraction algorithms are applied to label these pieces with music metadata (e.g., rhythm and melody). The mappings of the knowledge base are used to label music with affective metadata. This paper focus on the refinement of the knowledge base (subsets of features and their weights) according to the prediction results of listeners’ affective answers.