Seminar Abstract

29 Nov 2006:
Speaker:  Leandro Costalonga
Venue:     Devonport Lecture Theatre, Portland Square
Time:       14:00

Imperfections Simulation towards a Perfect Human's Guitar Performance

Computational models for music expressiveness and interpretation mainly consider parameters extracted from music performances compared with a previous-known score. This is known as the score matching technique.

This technique commonly extracts and feed the models with information about deviations in timing, tempo, dynamic (loudness) and articulation.
In the folk guitar context, the music notation commonly used is just tablature without rhythm information, this makes this strategy difficult to use.

The model that we propose goes beyond the analysis of the music performance; it also considers the biophysical constrains of the performer focusing in his pitfalls during the performance. We believe that a human's musical performance has mistakes and these imperfections contribute to expressiveness and interpretation of the music.