Descoberta automática de conhecimento em interpretações musicais: o caso do acompanhamento rítmico ao violão

It is common sense that playing music in the exact way it is written in the score results in a mechanical and uninteresting succession of sounds. In fact, the score is just a guide to the musician, who changes it according to his musical intentions, to some sort of empirical analysis of the musical structure and performance rules he learned during his musical life. These modifications are called expressive performance. Several researchers study different aspects of expressive performance. Results show that, although expressive performance is an art form, it still has common features across performers and performances. However, these results are restricted to Classical Music composed for the piano. Musical styles such as Bossa Nova and Samba and instruments like the acoustic guitar are rarely studied. In Computer Music, other researchers are concerned with the automatic analysis of expression. Some of them use techniques from fields such as Machine Learning and Data Mining in order to build models of musical expression. Nevertheless, they still analyze Classical Music composed for the piano. Despite all efforts, there still is a great number of unanswered questions about expressive performance. In this work, we explored a particular subset of these questions: Actually, how Bossa Nova’s rhythmic accompaniment is built by the performers. Given this new scope, we used a single unifying structural element, the rhythmic pattern, and established computational processes and algorithms whose results provided a better understanding of the phenomenon. The work resulted in the following contributions: an algorithm for the induction of the guitarist’s right hand fingering; a textual representation of the fingering, as well as a process for dimensionality reduction of this fingering; pattern extraction algorithm’s adaptations; and a prepared data set, as well as computational tools for the automatic rhythmic analysis.