Skip to Content

Department of Linguistics

MU-Talk: Speech Synthesisers

Robert Mannell

Macquarie University TTS systems preceding MU-Talk drove hardware-based or firmware-based speech synthesisers. These synthesisers included systems with purpose-built hardware, such as SID (developed by John Clark in the 1970s), or were based on code hosted on various firmware devices (such as a TMS320 signal processing chip-based system developed by a team led by John Clark in the 1980s, and a VLSI implementation developed by Clive Summerfield in the 1990s).

MU-Talk's TTS system drives software based speech synthesisers hosted on general purpose computers (currently on Windows PCs).

MU-Talk currently has a formant based diphone concatenation speech synthesiser loosely based on parallel formant firmware developed in our lab during the late 1980's and the 1990's. Another concatenation based synthesiser is also under development which uses diphones based on channel vocoded technology. Both of these implementations are awaiting the completion of processing (labelling and parameter extraction) of a new, already recorded, diphone database.

When time permits an already existing simulation of the old SID serial formant syntheser will also be incorporated into MU-Talk for pedagogic purposes. This system permits the running of synthesis-by-hand workshops which have proven to be useful in the past for teaching aspects of speech synthesis and speech acoustics.