FOF synthesis generates formant spectra from streams of grain bursts

Formant wave-function synthesis (FOF) generates a stream of grains separated by a time quantum corresponding to the period of a fundamental frequency. Each grain is a damped sine burst shaped by a specific envelope (the formant wave function). The grain’s duration and internal frequency determine the formant peak location and bandwidth. Multiple simultaneous FOF streams, each tuned to a different formant, can synthesize vowel sounds with high quality. FOF was developed at IRCAM (Rodet, Potard, Barriere 1984) as a highly efficient method for vocal synthesis. It produces convincingly natural vowels with far fewer computations than additive synthesis.

Examples

FOF voice: three streams at formant frequencies 700 Hz (F1), 1200 Hz (F2), 2700 Hz (F3), fundamental = 120 Hz. The combined streams produce a sung vowel /a/.

Assessment

What is the relationship between the time interval between FOF grains and the perceived fundamental frequency? How does changing the grain’s internal frequency affect the output spectrum?