Formant-Based Audio Synthesis Using Nonlinear Distortion
Miller Puckette · 1995
An audio synthesis technique ("Phase Aligned Formant synthesis") is presented which is aimed at real-time musical applications. A desired sound is specified in terms of one or several time-varying formants, each with specified center frequencies, bandwidths, and amplitudes. The sound produced may be periodic or noisy. The relative merits of other known real-time techniques for synthesizing sounds with desired formants are also discussed. As an example, a spoken word is analyzed and resynthesized. 1 Introduction Designing an instrumental voice for a synthesizer often involves planning how the various partials of the sound will vary in frequency and amplitude over the life of a note. In many natural sounds the timbre is best described in terms of the spectral envelope, which gives a pitch-independent specification of the evolution of a sound's spectrum in time. This is especially true of the spoken or sung voice, whose spectral evolution determines (among other things) the phoneme or p...