In this thesis, a new method of realization of a digital audio response system has been investigated. The residual-excited linear prediction (RELP) vocoder with the pitch predictive loop has been used as the speech digitization method. The hardware used to implement this system is a "bit-slice" microprocessor based microprogrammed system for the real-time signal processing. Except for analog postprocessing the entire system is all digital and has been developed by software.
Throughout the thesis a particular emphasis has been placed on getting the best quality of synthetic speech. As a result we believe that speech quality of this voice response system is far superior to any other available speech synthesizer.