Intelligent Genetic Fuzzy Inference System for Speech Recognition: An Approach from Low Order Feature Based on Discrete Cosine Transform

摘要

In this paper an intelligent methodology for speech recognition, is proposed. In addition to processing, with mel-frequency cepstral coefficients, the discrete cosine transform is used to generate a two-dimensional time matrix for each pattern to be recognized. A Mamdani fuzzy inference recognition system is optimized by genetic algorithm to maximize the hits of patterns with minimum number of encoding parameters. Experimental results for digit recognition applied to Brazilian language show the efficiency of the proposed methodology compared to others techniques widely cited in the literature...