A modularized speech recognition processor LSI with a highly parallel structure

To actualize a continuous speech recognition System with a large vocabulary, we proposed ring-array-processor architecture [1-2], This architecture has the two features: highly parallel DTW(Dynamize Time Warping) processing [31 capability. which is the main algorithm used to realize speech recognition. and array size flexibility. which makes it possible to determine the number of PE(processing Element) contained by the array processor according to vocabulary size.