A speech recognition system using linear predictive coding and dynamic time warping

A system capable of recognizing isolated speech utterances with a rate of 97% and rejection rate of 2% on a 200-wood vocabulary is presented. The system uses linear predictive coding and dynamic time warping, both implemented in software on an Apple Macintosh computer. The Macintosh also provides an environment for a graphics human interface in the authors' speech research. The system is intended for vocal shaping of severely disabled individuals.<
>