Research And Development Of English Learning App System Based On Neural Network

With the advent of the Internet era,mobile applications are becoming more and more popular.Autonomous and customized mobile learning methods are more convenient than traditional classroom learning methods.The development and application of artificial intelligence technology also make mobile learning methods more accurate and efficient.Speech interaction always has been the focus of human-computer interaction breakthroughs.It includes many technologies such as speech recognition,natural language processing,and speech synthesis.These technologies can use neural networks to train reliable models,which are very suitable for language Used in teaching.Aiming at the lack of oral training in domestic mobile English learning applications,this article has developed an English learning app based on speech recognition and synthesis technology and neural network technology.The app developed in this article is developed using Flutter,g RPC,Tensor Flow,and other technologies.The front-end includes the user center,shopping mall,and voice interaction functions,and the back-end includes course management,payment,and neural network model training functions.The main work of this paper includes:1)Propose the overall architecture design plan of the English learning app system based on neural networks.Through demand analysis,the business functions are sorted out and the system modules are divided.Its main functions include user management,course management,shopping cart,payment,speech recognition,and speech synthesis.2)Give a mobile client architecture scheme based on Flutter,use the MVC design method to layer the mobile terminal structure,the upper layer is the UI layer,the middle business layer,the lower layer is the model layer,mobile devices have microphones and other devices through the tool layer to use the Flutter SDK Communicate with mobile devices.3)Give a g RPC-based server system design plan,provide mobile clients with data services such as users,courses,orders,neural network models,etc.,and build a payment refund service by using a third-party payment interface.4)Propose a neural network training system using Tensor Flow to train the speech recognition model and speech synthesis model,and use the migration module to compress and migrate the trained neural network model to the mobile terminal for operation.5)Set up a test environment,use Nginx and Docker to configure the deployment server,and use an end-to-end test method to test the system functions.The test covers the client,server,and database.The neural network model evaluation uses word error rate and MOS scoring mechanism to test speech recognition and speech synthesis functions respectively.The test process adopts a multi-group comparison method,and finally,the test results are given.The application has been tested and can meet the learning needs of learners.