Composing with EncycloSpace: a Recombinant Sample-based Algorithmic Composition Framework

Author:
Spitsyn, Yury, Music - Graduate School of Arts and Sciences, University of Virginia
Advisor:
Shatin, Judith, Department of Music, University of Virginia
Abstract:

Recorded sounds have been in compositional circulation since the early days of musique concrete, yet there have been major lacunae in the design of computer tools for sample-based algorithmic composition. These are revealed by the absence of algorithmic methods based on quantified representation of the sampled sound’s content. The parametric utilization of recorded sounds lags far behind that of synthetic sound methods, which are parametric by design.
During the last decade, musical informatics have established computational methods to break through the informational opaqueness of sound recordings. By extracting semantic data (called descriptors or features), sampled sounds can be converted into quantifiable form and more easily described, diagrammed and compared. This technology has resulted in numerous compositional applications focused on descriptor-based sound manipulations and database usage. Surprisingly, compositional strategies have gravitated to only a narrow range of manipulations. Inherent limitations in current approaches motivated a search for open-ended design principles for corpus-based sample-oriented algorithmic composition.
This dissertation examines theoretical and practical ways to approach this problem before proposing a solution. To avoid conceptual shortcomings, we start with an overarching perspective by introducing the ontological topology of music as developed by Guerino Mazzola. We use his comprehensive analysis of activities, scientific disciplines, ontological dimensions and informational processes in music as a basis for our approach to system design in a balanced and deliberate way. Mazzola's notion of EncycloSpace, the knowledge base of musical process, and the concepts of receptive and productive navigation serve as a conceptual core for the design. The overall system's structure is tripartite: an analytic subsystem, a compositional subsystem and EncycloSpace - database subsystem.
A methodology, termed NOTA-transform, is proposed as a general method for the compositional use of EncycloSpace. It defines four consecutive transformations, called navigation, ordering, temporalization and adaptation, as indispensable stages of converting EncycloSpace into a musical object. Each of the stages is defined as an independent decision-making site.
The system design considerations are then used to implement a working software prototype called sEl, described in detail in Chapter 3.
Finally, sEl is utilized to compose a piece, Ignis Fatuus, where I demonstrate the resulting analytic and compositional algorithms.

Degree:
PHD (Doctor of Philosophy)
Keywords:
algorithmic composition framework, algorithmic composition, recombinant composition, corpus-based composition, computer-aided algorithmic composition
Language:
English
Rights:
All rights reserved (no additional license for public reuse)
Issued Date:
2015/07/20