Speech Recognition Method Based on Normalized Simplified Artificial Fish Swarm Algorithm

Abstract

In order to improve the classification performance of the speech recognition system, aiming at the problem that the traditional artificial fish swarm algorithm runs the late search blindness, low optimization precision and slow calculation speed, the artificial fish swarm algorithm is simplified by a certain method. A speech recognition method based on normalized simplified artificial fish swarm algorithm is proposed according to a relationship between data structure and optimization algorithm. Firstly, the speech features extracted by Mel Frequency Cepstrum Coefficient is normalized to reduce the complexity of the data structure. Secondly, the streamline operation is used to improve the iterative process of artificial fish swarm algorithm, and the modified algorithm is applied to the support vector machine parameter optimization of speech recognition. Finally, the optimized support vector machine model is used to classify and identify the normalized speech features. The experimental results show that the proposed algorithm improves the speech recognition rate by8.58% in average compared with the traditional artificial fish swarm algorithm, and the maximum increase of 15.34%, and has good anti-noise performance and generalization ability under high signal to noise ratio and large vocabulary.