SPL: A speech synthesis programming language

Abstract

This report describes the first version of a high level computer programming language for experiments with synthetic speech. In SLP a context sensitive parser is programmed to recognize linguistic constructs in an input string. Both the structural and phonetic descriptions of the recognized structures may be modified under program control. The final output of an SPL program is a data stream capable of driving a parametric speech synthesizer. The notation used is based on the principles known from Chomsky and Halle's "The Sound Pattern of English". This means that in principle all linguistic constructs are programmed in segmental units. However, in SPL certain macro facilities have been provided for more complicated units such as syllables or words.