Lexical prediction for a text-to-speech system

A lexical prediction system is being developed at the Department of Speech Communication and Music Acoustics to be used with our speech synthesizer (Carlson, Granstrom and Hunnicutt, 1982). It is being developed in response to the observation that some non-speaking users of the synthesizer find communication with such an aid laboriously slow. Cme user commented that the other children were guessing the words she was typing before she finished, but that she had to finish typing them anyway for the speech to be correct. The lexical prediction system addresses this problem. It is implemented on an Eclipse computer in a test system with speech output from a connected synthesizer. Hardware implementation has begun while the system undergoes continued development and testing.