Spectro-temporal control parameters for synthetic orchestra applications

Abstract

Several spectral properties can be exploited for control of individual voices in a synthetic orchestra. Sounds can be synthesized from their spectral representations using time-varying additive synthesis, inverse Fourier transform, or, more efficiently, using multiple wavetable synthesis. Primary parameters to be controlled are amplitude and spectral centroid versus time envelopes, spectral envelope, spectral irregularity, vibrato, noise content, and inharmonicity. These parameters can be further resolved into subparameters. For example, amplitude versus time curves can be partitioned into attack, steady-state, and decay segments which can be manipulated separately. For sustained sounds, it is also important to consider how these parameters vary with fundamental frequency and intensity. In the case of percussion sounds, the density of partials as well as their spectral evolution are important. In all cases, detailed management of parameters is necessary to maintain timbral clarity, uniqueness, and naturalness. Examples of synthetic orchestral instruments will be played.