Modeling Expressive Music Performance in Bassoon Audio Recordings Rafael Ramirez, Emilia Gomez, Veronica Vicente, Montserrat Puiggros, Amaury Hazan, and Esteban Maestre Music Technology Group Pompeu Fabra University Ocata 1, 08003 Barcelona, Spain Tel:+34 935422165, Fax:+34 935422202 {rafael,vicente,puiggross,hazan,maestre,gomez}@iua.upf.es Abstract. In this paper, we describe an approach to inducing an ex- pressive music performance model from a set of audio recordings of XVIII century bassoon pieces. We use a melodic transcription system which ex- tracts a set of acoustic features from the recordings producing a melodic representation of the expressive performance played by the musician. We apply a machine learning techniques to this representation in order to induce a model of expressive performance. We use the model for both understanding and generating expressive music performances. 1 Introduction Expressive performance is an important issue in music which has been studied from different perspectives (e.g. [2]). The main approaches to empirically study expressive performance have been based on statistical analysis (e.g. [11]), math- ematical modelling (e.g. [13]), and analysis-by-synthesis (e.g. [1]). In all these approaches, it is a person who is responsible for devising a theory or mathemat- ical model which captures different aspects of musical expressive performance. The theory or model is later tested on real performance data in order to deter- mine its accuracy. In this paper we describe an approach to investigate musical expressive perfor- mance based on machine learning [7]. Instead of manually modelling expressive performance and testing the model on real musical data, we let a computer use an inductive logic programming algorithm to automatically discover regulari- ties and performance principles from real performance data (i.e. bassoon audio performances). The rest of the paper is organized as follows: Section 2 describes how the acoustic features are extracted from the monophonic recordings. In Section 3 our approach for learning rules of expressive music performance is described. Section 4 reports on related work, and finally Section 5 presents some conclusions and indicates some areas of future research. D.-S. Huang, K. Li, and G.W. Irwin (Eds.): ICIC 2006, LNCIS 345, pp. 951–957, 2006. c  Springer-Verlag Berlin Heidelberg 2006