The paper describes a Transputer-based English text-to-speech system which runs in quasi-real time, is efficient in data memory, non-demanding in labour and is capable of text to speech transcription of high accuracy. The system described is multialgorithmic in that three algorithms representing different approaches to transcription are executed concurrently. Intercommunications and arbitration between algorithms result in sound phonemes which are of acceptable accuracy with respect to character vowel, length and intonation.