Fuzzy speech recognition

Speech recognition is a major topic in speech signal processing. Many algorithms based on results of speech analysis, among which dynamic time warping) and hidden Markov models are the most important, have been advanced. However, these algorithms generally turn out to be too complicated to be implemented in real time systems. The proposed algorithm in this paper, which uses a fuzzy logic recognition approach based on the power distribution pattern of a segment of a speech, allows the implementation of real-time speech recognition