A modular audio response system for computer output

We describe a flexible, modular system for representation of a text-to-speech system which provides for ease of understanding, facile development, a variety of end-use systems, exportability, and ready implementation into novel technology to meet real-time needs. This system thus serves as a research base, a framework for experimentation, and the seat of structural formalisms which can imply special hardware processors. Since the relation of the surface speech waveform to the underlying linguistic description is becoming increasingly complex, we expect that the virtues of this approach will become necessities rather than conveniences.