The Application of Neural Network in the Voice Signal Processing System

摘要

This paper adopts the genetic algorithm to set up a model of neural network to resolve the fuzzy problem in the fault diagnosis. In this paper, a new combining algorithm-GAL algorithm is proposed, which combines the genetic algorithm with BP algorithms. Then the design of neural networks classifier is proposed. To capture the cepstrum feature of voice, an efficient normalization that combined biggest-smallest method with sqrt method is proposed. The results show that the ratio of recognition of the voice signal processing system based on the GAL algorithm proposed is better than the traditional methods.