Study and development of symbolic and acoustic descriptors for the analysis, composition and musical performance

• Grant number: 10/06743-7
• Support Opportunities: Regular Research Grants
• Duration: January 01, 2011 - December 31, 2012
• Field of knowledge: Engineering - Electrical Engineering
• Principal Investigator: José Eduardo Fornari Novo Junior
• Grantee: José Eduardo Fornari Novo Junior
• Host Institution: Núcleo Interdisciplinar de Comunicação Sonora (NICS). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil

Abstract

Literature of MIR (Music Information Retrieval); the interdisciplinary science for the studying and collecting of musical aspects, involving: engineering, neuroscience and musicology, among others, defines "music descriptor" as a computational model capable of predicting - to a certain extent - musical aspects (categoric, scalars or time series) emulating human auditory (perceptual and cognitive) capacity of its identification. A musical aspect is a singular feature of human auditory sensation, brought by the presence and variation of intensity of its partial components, in the time domain (i.e. rhythmic aspects, pulsation, dynamics, etc.) and frequency domain (i.e. melodic and harmonic aspects, timbre, etc.). Symbolic descriptors predict musical aspects by collecting parametric data from music notation (i.e. music scores, MIDI files, etc.). Acoustic descriptors retrieve musical aspects data from audio files. Music, as an artistic expression, has three categories of human interaction: Analysis (the study aiming to understand the rationale behind the structuring of musical compositions), Composition (the creation and notation of musical structures), and Performance (the manifestation of musical structures throughout sonic information). In spite of the existence of several developments of computational models for the acquisition of musical aspects, their results are still partial, sometimes redundant and often inconsistent; most likely due to the fact that MIR; a relatively new interdisciplinary field, has many studies conducted by researchers from different fields of science. This project aims to continue the research developed by the author during his PosDoc at the European project: Braintuning (www.braintuning.fi ) where he developed algorithms for musical aspect acquisition that achieved great efficiency [26,27,28,29]. The goal of the project here proposed is to: study, classify, and possibly develop new music descriptors and to apply them into musical processes of analysis, composition and performance. (AU)