Static Analysis of Performance Parameters for Music Scores

Context : Music performance is by nature non deterministic : it is well known that two interpretations of the same music score, where all the notes and their durations are specified, can differ significantly, in particular wrt temporal aspects. However, despite the temporal variability, several musicians playing together manage (in general) to make the music sound " as expected " , using various synchronization strategies (mostly implicit). In the case of mixed instrumental/electronic music, where a computer has to play together with musicians in realtime, the synchronization strategies must be formally specified, which is not an easy task (see e.g.[6]). The goal of the internship it to help composers in specifying interaction in mixed music by providing them with procedures for the static analysis of the temporal behavior of written scores. Antescofo [1] is a system used in performance for interpreting electronic music scores in realtime synchronization with a live musician. It is capable of following the performance of the musician, decoding in realtime its position within its score and its current tempo (pace) [4], and of synchronizing this performance with computer realized elements. The music scores currently used in Antescofo, consist essentially in the specification of sequences of musical events to be recognized (from the performer audio stream in input) and of electronic actions to be output by the computer in response. Various strategies are also proposed for specifying synchronization between events and actions and for error handling [5]. These specifications can be seen as a special kind of timed systems where the delays (between an event and an action or between two successive actions) are expressed relatively to the recognized tempo (which changes at each element detected).