Modelling the vocal tract using multiple digital signal processors

Describes the experience obtained during the design, build and operation of a multiple digital signal processing system based on Texas Instruments TMS32020 chips. The system was designed specifically to explore the use of an electrical transmission line model representing the operation of the human vocal tract in which an utterance is defined by a set of parameters indicating the way in which the model is changing. The architecture adopted is that of a master 32020 communicating with (presently five) slave 32020s through shared global data memory. Each slave operates with local program memory (ROM and RAM) and the on-chip data RAM. The master communicates with a host personal computer through shared data and program memories.<
>