Research on the application of biomimetic computing in speech recognition

Since speech is one of the most direct and effective means of human communication, itpsilas natural to apply biomimetic processing mechanism to automatic speech recognition to solve the existing speech recognition problems. In this paper, three typical techniques were selected respectively: simulated evolutionary computation (SEC), artificial neural network (ANN) and fuzzy logic and reasoning technique, from intelligence building processing simulation, intelligence structure simulation and intelligence behavior simulation, to identify their applications in different stages of speech recognition. A SEC usage and the corresponding algorithm were presented to generate training speech data, and a method of Elman network as acoustic model was also presented.