Programmable, digital speech synthesiser

Abstract

A low-cost, programmable, digital speech synthesiser, to be used for research purposes, is described. This machine has a general-purpose architecture of arithmetic unit, memory and control unit, which enables it to simulate formant-type synthesisers as well as vocal-tract models, all in real time. The principle virtue of the machine is the ease with which changes in the structure of a synthesiser can be evaluated, thus facilitating the rapid optimisation of a proposed design.