Paradigmatic analysis using genetic programming

• Author(s): Grilo, Carlos Fernando Almeida ; Machado, Fernando ; Cardoso, Fernando Amílcar Bandeira
• Date: 2001
• Persistent ID: http://hdl.handle.net/10400.8/87
• Origin: IC-online
• Subject(s): Musical piece; Paradigmatic analysis; Genetic programming
• Document Type: Conference Object
• Language: English

Description

Paradigmatic analysis consists in the segmentation of a musical piece through the identification of relations between different parts of the piece, and the classification of the identified segments into categories. In this paper we describe how a genetic programming system can be used to make the paradigmatic analysis of monophonic musical pieces, using a simple fitness function inspired in the Kolmogorov complexity estimation. We make use of automatically defined functions in order to represent segments. Relations are made explicit through the reuse of segments and the application of transformations to these segments.