SPASS—An Improved Terminal-Analog Speech Synthesizer

Abstract

A new terminal analog or resonance speech synthesizer has been developed combining analog and digital computation techniques. The signal-processing speed of analog circuits is merged with the symbolic and arithmetic manipulation flexibility of digital computers to provide performance surpassing that possible with strictly analog, or strictly digital techniques. The familiar cascaded-pole synthesizer is realized by ordinary analog-computer methods, with circuit coefficients determined by digital multipliers controlled from a small digital computer. The experimenter exercises control over the synthesizer with a series of commands typed on a typewriter console. By this method, an utterance may be synthesized, judged, modified, and tried again in rapid succession. The data specifying an utterance may be entered from prepared paper tapes, by fitting polynomials to points entered through the typewriter, or by drawing the required data with a “light pen” on a CRT display. An algebraic statement interpreter is included to process input data and facilitate generating regular utterance sequences. [Work supported in part by the U. S. Air Force Cambridge Research Laboratories.]