Research On Voiceprint Recognition Algorithm Based On Deep Neural Network

With the development of society and the continuous progress of communication and information processing technology,biometric technology and people’s information security are more and more closely linked,the traditional identity authentication method has been unable to meet the needs of people for their own information security.As a kind of biometric technology,voiceprint recognition has attracted more and more attention and has been widely used in mobile payment,intelligent terminal,medical service and criminal investigation.However,due to the short research of voiceprint recognition,there are still many problems to be solved.When extracting the voiceprint features of a speaker,the noise will distort the speaker’s speech spectrum.The traditional natural language processing model is composed of several steps,each step is an independent task,and its results will affect the next step,thus affecting the final recognition results.The purpose of this study is to study the method of Voiceprint Recognition Based on deep learning system.Through the analysis of speech enhancement algorithm of speech front end,it is concluded that RLS algorithm is more suitable for speech enhancement.Based on the traditional RLS algorithm,this paper proposes a new algorithm based on RLS,which is based on the traditional RLS algorithm.The experimental results show that the Improved RLS algorithm can suppress the effect of noise on feature extraction.Aiming at the problem that the recognition accuracy is reduced due to the error accumulation caused by multi-step and multi module in the current speech processing process,the improved GE2 E model is proposed to optimize the feature extraction after comparing and analyzing the comprehensive effects of several endto-end models.Based on the GE2 E model,the algorithm optimizes the deep neural network model to make better use of the context of each node in the input layer and output layer,and reduces the complexity of the GE2 E model.The experimental results show that the improved GE2 E model can improve the recognition performance of the model and greatly reduce the training time.