FOF synthesis is a technique for producing waveforms with a specific spectral shape. It was developed by Xavier Rodet at IRCAM. Similar to the VOSIM technique, sounds are produced from a sequence of excitation pulses in the form of enveloped sine waves. The rate of generation of the excitations specifies the fundamental frequency of a formant region. Harmonics of this fundamental are also present, and their amplitude is determined by the shape of the envelope that is applied to each sinusoid in formaing the excitation pulses. The shorter this envelope, the more gradual the falloff in harmonic amplitude in the frequency domain. Lengthening the envelope, on the other hand, reduces the harmonic amplitudes quickly in the the frequency domain, narrowing the width of the formant region. Other details of the local envelope shape, such as the rise and fall rates, affect the shape of the harmonic amplitude curves in subtle ways, allowing for precise control of the harmonic structure. Of course, the shape of the local envelope can be changed over time, permitting dynamic timbres to be produced. Typically, many such excitation waveforms are added together, each providing a different formant region. In this way vowel-like vocal sounds can easily be created, although non-vocal sounds are also possible.
Taken from:
http://www.cim.mcgill.ca/%7Eclark/nordm ... r.html#fof