Answers are grounded in this paper and cite their sources.
A relatively small computer program produces synthetic speech from English text for a phoneme-driven synthesizer. The speech is not inflected, but within that constraint is deemed intelligible on at least 97% of the words in typical runing text. The program works almost entirely by rules, most of which have a stereotyped form. Since no large dictionary is needed, secondary storage is not used. All the rules are given, together with examples of their behavior.
No takes yet. Share an insight, caveat, or question.
M. D. McIlroy (1974) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: