Companhia Não-Humana: Praticando Improvisação Livre Pela Interação Com Máquinas

Abstract

Free improvisation is a musical genre that challenges traditional ways of making music and is generally defined by its absences. This creates a particular environment that poses challenges for those who wish to learn it, and for educators inspired to teach it. Generally, ensemble practice is considered the most suitable and widely accepted method for getting started in the genre. However, should it be the only one? This text discusses a contextualization of the practice of free improvisation in relation to other cultures of improvised music, looking at different approaches to teaching it and establishing bridges between the theory and practice of the genre. Here we propose the use of musical agents as a possible alternative for the individual practice of free improvisation. To this end, we developed an open-source improvising machine in the Pure Data visual programming language, named “engenhoca”. The system is able to listen to musical elements of the sound events played by a musician with whom it interacts in real time, classifying and grouping them into musical phrases to be stored in its memory. These phrases are eventually drawn in response to changes identified in the material played by the musician. The different elements of these phrases are modified by mathematical operations and sent to a module that synthesizes electronic sounds in real time with the data received. engenhoca was evaluated in a non-experimental way, collecting quantitative and qualitative data from experienced musicians (n=3), who were invited to play with the system for three improvisation sessions and express their impressions on different aspects of the interaction. We believe that the project is compatible with the aesthetics found in free improvisation and represents a plausible approach to musical practice in this context.