Computational Analysis of Expressivity in Classical Guitar Performances

Xavier Serra
Josep Lluis Arcos

Co-advisor: Xavier Serra, UPF

Data lectura: 29/11/2013

The study of musical expressivity is an active field in sound and music computing. The research interest comes from different motivations: to understand or model musical expressivity; to identify the expressive resources that characterize an instrument, musical genre, or performer; or to build synthesis systems able to play expressively. To tackle this broad problem, researchers focus on specific instruments and/or musical styles. Hence, in this thesis we focused on the analysis of the expressivity in classical guitar and our aim is to model the use of expressive resources of the instrument.
The foundations of all the methods used in this dissertation are based on techniques from the fields of information retrieval, machine learning, and signal processing. We combine several state of the art analysis algorithms in order to deal with modeling the use of the expressive resources.
Classical guitar is an instrument characterized by the diversity of its timbral possibilities. Professional guitarists are able to convey a lot of nuances when playing a musical piece. This specific characteristic of classical guitar makes the expressive analysis is a challenging task.
The research conducted focuses on two different issues related to musical expressivity. First, it proposes a tool able to automatically identify expressive resources such as legato, glissando, and vibrato, in commercial guitar recordings.
Second, we conducted a comprehensive analysis of timing deviations in classical guitar. Timing variations are perhaps the most important ones: they are fundamental for expressive performance and a key ingredient for conferring a human-like quality to machine-based music renditions. However, the nature of such variations is still an open research question, with diverse theories that indicate a multi-dimensional phenomenon. Our system exploits feature extraction and machine learning techniques. Classification accuracies show that timing deviations are accurate predictors of the corresponding piece.
To sum up, this dissertation contributes to the field of expressive analysis by providing, an automatic expressive articulation model and a musical piece prediction system by using timing deviations. Most importantly, it analyzes the behavior of proposed models by using commercial recordings.

Otras publicaciones del autor