Rules for demisyllable synthesis using Lingua, a language interpreter

A speech synthesis system, called Lingua, has been developed to generate speech from demisyllables. The set of programs permits features and units to be defined by the user, and permits feature values to change asynchronously within the unit. Rules for determining duration, amplitude, pitch, etc. are specified in a format based on generally used linguistic notation. The inventory of units for the synthesis comprises fewer than 1000 LPC-encoded demisyllables previously generated by Lovins et al. (see Lovins, Macchi, and Fujimura, JASA 65, SI, Spring 1979).