Experimental speech input/output system

The authors present the hardware and software design of an experimental system allowing the inclusion of speech capabilities at the man/machine interface level. The hardware is based on a personal computer in combination with a multiprocessor system. The latter consists of a microprocessor for management purposes and several signal processing boards for the implementation of speech algorithms. An audio board provides the interface to the acoustical environment. The software consists of a high-level supervisor function, a medium-level hardware manager, and low-level functions for communication between the different DSPs (digital signal processors) and the system. Speech modules (speech recognition, speech coding, speaker verification, and text-to-speech synthesis) are also implemented