Application of the neural ring pattern classifier to speech recognition
Pyeatt, Larry D.
MetadataShow full item record
The Neural Ring is based on the avalanche neural network model which was proposed by Grossberg. The avalanche model was modified by Hecht-Nielsen to form a Nearest Matched Filter classifier. The Neural Ring, suggested by Taber, uses the Hecht-Nielsen Nearest Matched Filter classifier, but also incorporates an activation latch and fuzzy logic to increase classification performance. This paper describes research in which the Neural Ring model is evaluated as a tool for human speech recognition. Recognition results were tested on a ten element vocabulary consisting of the words 'zero' through 'nine' as spoken by several different speakers.