Introduction to the IPEM Toolbox for Perception-based Music Analysis

The motivation for developing a toolbox for perception-based music analysis starts from the observation that much of the world music production (seen over different cultures and time periods) has no score representation, and that if a score representation is available, it still represents but a small percentage of what musical communication really is about. A sonological analysis of musical sounds may provide steps towards a better understanding of the components that contribute to musical information processing but is also insufficient in view of how humans deal with musical signals. Technology has revolutionized our conception of what kind of music research will be possible in the 21st Century. But it is the task of musicologists, in collaboration with other scientists, such as engineers, psychologists, and brain scientists, to make these dreams more concrete. Our basic motivation for developing this toolbox is a pragmatic one: if the musicology aims at understanding music as a social and cultural phenomenon embedded in the physical world and mediated through human faculties of perception, cognition, and processing of expressive communication, then new tools must be developed that allow a fully integrated approach. This paper introduces the main concepts and a state-of-the-art of the modules of the IPEM toolbox for perception-based music analysis. This toolbox represents a bottom-up approach to musical description taking human perception as the basis for musical feature extraction and higher-level conceptualization and representation. The IPEM toolbox provides a collection of MATLAB functions that allow dealing with different aspects of feature extraction as well as a global concept concerning perception-based music analysis. First we present the global internal representation framework. Then we sketch what we have in mind as a global picture and present the modules currently involved. Finally we describe some applications of the toolbox.