Comunicació presentada al International Symposium on Performance Science (ISPS), celebrat a Viena (Àustria) els dies 28 a 31 d'agost de 2013.

We present an exploratory study on ensemble expressive performance based on the analysis of string quartet recordings. We recorded a piece with three expressive intentions: mechanical, normal, and exaggerated. We made use of bowing gesture data (bow velocity and force) acquired through a motion tracking system to devise a precise score performance alignment. Individual contact microphone audio signals allowed extraction of a set of audio descriptors for each musician and each note. We show how tempo and loudness on a macro-scale changed across expressive intentions and score sections. The score is also taken into account in the analysis by extracting contextual attributes for each note. We show that micro-deviations were affected by note contextual attributes, whereas the effect of expressive intention varied across sections. We find sections that exhibited a lower entrainment, where individual parts tended to be freer and presented more asynchronies.

This work was supported by EU FET-Open SIEMPRE and by SIEMPRE-MAS4.