Excitation of Vocal-Tract Synthesizers

A method is described for representing voiced and voiceless excitation in the vocal tract. Three physiological factors suffice for synthesis of nonnasal speech: subglottal pressure, vocal-cord tension, and vocal-tract shape. Voiced excitation is obtained from an oscillator model of the cords. Voiceless excitation is produced by elements that approximate turbulent flow in the tract. These elements are controlled automatically by factors related to the acoustic volume velocity in the tract and the tract shape. Voiced, voiceless, and voiced-fricative sounds are synthesized from a vocal tract incorporating these sources. The whole system is computer simulated.