FOF (Formant Wave Function Synthesis) was developed as part of IRCAM's 'CHANT' project, which was devoted to the development of techniques for the realistic synthesis of the singing voice. It models the human voice as the sound from an impulse generator (the equivalent to the vocal chords) passing through a set of band-pass filters (representing the characteristics of the vocal tract), each filter corresponding to a vocal formant.
Since the output from the impulse generator can be considered as a sequence of 'grains', the technique can be seen to be closely related to granular synthesis, the difference being that in FOF the grains are regular and synchronous, generating a coherent periodic waveform. (Source - Richard Dobson (1992). A Dictionary of Electronic and Computer Music Technology. Oxford University Press.)