Parametric Artificial Talking Device

Abstract

The Parametric Artificial Talking device is described. Earlier forms of this apparatus for speech synthesis were demonstrated to this Society in 1953 and 1956. The speech synthesis circuits are designed as analog computers and can be set up to simulate various mathematical models for the system function and the excitation. The parameters of these functions can be varied by changing the values of control voltages, and, when these are suitably varied, complete words and phrases can be synthesized. A new controller has been developed which allows up to 12 separate parameter control functions to be generated simultaneously. This instrument is described in a companion paper. In one mode of operation the outputs of the formant generators are combined in parallel. In this form the device has been used to synthesize speech from spectrographic analysis and to show intelligible speech can be synthesized from a phonemic input by rules. This work is described in a companion paper. In another mode of operation the formant generating circuits are connected in tandem, conforming to the pole-zero mathematical model developed by Fant. The positions of the poles and zeros can be related by acoustic theory to the shape of the vocal tract, and the excitation can be related to vocal cord movements, enabling intelligible speech to be synthesized from a knowledge of the details of articulation.