On-Line, Real-Time Multiple Speech Output System and Its System Evaluation.

Abstract

A new multiple speech output system is described. This system is based on a compilation method. Fundamental speech elements used for the synthesis are the speech waves within one cycle at larynx frequency. They are stored on a secondary memory such as a magnetic drum or disk in the form of digitalized zero-crossing intervals. Speech elements are read out from the secondary memory and are connected in succession according to the connection rules by the computer program. The resulting sequences are transformed into the speech sounds by synthesizers. System evaluation is also performed in real time mode to measure the load of the computer and the maximum number of multiplicity.