Parametric Representations

Another class of representations has nothing to do with models of hearing or acoustics in particular, but exploits some numerical properties to achieve high-quality audio results. The first major breakthrough in this arena was frequency modulation, developed at Stanford by John Chowning [42,43]. FM had been used for several decades for radio transmission. Chowning discovered that certain combinations of carrier and modulator frequencies could produce waveforms that mimicked the properties of musical waveforms. FM was used in digital synthesisers and other musical devices, mostly by Yamaha. Many PC plug-in cards also use FM chips for sound synthesis.
Other techniques such as granular synthesis, waveshaping, Chant, or the Karplus-Strong algorithm have been investigated in the computer music community and implemented in some synthesisers such as the Korg ½ Series (waveshaping). Techniques in the speech community such as vector quantisation are discussed below.