Neural networks and radial basis functions in classifying static speech patterns

摘要: Abstract This paper compares the performances of three non-linear pattern classifiers in recognition static speech patterns. Two these are neural networks (Multi-layered perceptron and modified Kanerva model). The third is method Radial Basis Functions. A review several classification techniques similar to radial basis functions presented. class boundaries generated by different methods compared on simple two-dimensional examples. Experiments classifying eight vowels from a subset DARPA TIMIT database reported.