Database-Driven Systems and Applications for Music Information Management

The aim of this document is to explore the singing voice as a valid musical controller for real-time synthesis. We address the identification and extraction of the voice features, and the mapping of those features to a synthesis engine. Since the advent of digital music synthesis and MIDI, many controllers have been developed, ranging from the traditional keyboard to sophisticated computer vision systems. We propose to employ an ubiquitous and highly expressive instrument the voice. This work first reviews aspects of voice analysis, and the control of different synthesis techniques. Finally, we present a prototype of {\em Vocal Controller} that synthesizes a bass guitar using two techniques Physical Models and Spectral Models. The proposal for the PhD dissertation consists of extending the current prototype, allowing for real-time performance situations; and validating the system with two groups of subjects naive users and skilled musicians