Speech Recognition by Classifying Speech Signals Based on the Fire Fly and Fuzzy

In this paper, a fuzzy speech recognition system is proposed for classifying speech signals to get more accurate recognition with higher speed. This system is combined with 5-layers fuzzy logic and is more accurate in speech recognition than other methods like: particle swarm optimization-Forward Neural Network (PSO-FNN) and Back Propagation Forward Neural Network (BP-FNN). In this paper, speech samples are first given to input of the fuzzy circuit to check (investigate) signals in a fuzzy framework and a pattern of signals is produced for each signal cluster. This causes dimension reduction of signal data and gives us better and more reliable recognition result. For recognizing speech, we use firefly classification method and consider a special class for each input to improve recognition rate. Classifying fuzzy signal is the reason for increasing recognition accuracy. Our method is also capable of recognizing noises in environment around and consider each noise individually as a cluster and then removing it from input signal for final recognition. Our classification method based on firefly algorithm improves recognition of speech signals in the proposed model.