Design And Implementation Of Table Tennis Training System Based On Intelligent Voice Interaction

In recent years,with the rapid development of artificial intelligence technology,the application field of intelligent voice interaction platform has been further expanded.At present,it has been widely used in intelligent inquiry call center,voice interaction system in smart home,vehicle voice interaction system,voice assistant in smart phone and intelligent voice dialogue of government robot,etc.The shadow of voice interaction equipment can be seen everywhere in our life.The research of applying artificial intelligence technology to the entertainment system has attracted great attention and become one of the research hotspots in this field.This paper takes table tennis intelligent training system as the research object,and focuses on the research of table tennis intelligent voice interaction system,integrates voice interaction and natural language processing technology,and constructs intelligent table tennis training scene through reasonable system design,aiming to provide an intelligent training system for table tennis fans.The system platform makes corresponding operations through face recognition,determining the identity of training personnel,receiving voice instructions from specific users.The main functions include playing and controlling the relevant teaching video according to the training plan set by the system platform coach,controlling the service mode of the table tennis server by coach plan,recording and playing back the monitoring video,answering the user's questions professionally,inspecting the training and making the training plan for the students.The intelligent Q & a module of this system is realized by retrieval and generative.For generative Q & a module,the skip-gram model in word2 vec is used to get the word vector,the LSTM is used as the neuron of training network,and the generative Q & a system is realized based on the seq2 seq model.In order to solve the problem that the information in the back part of the sequence will cover the information in the front part of the sequence in seq2 seq,In the training model,the attention mechanism is added to the encoder module of seq2 seq to allocate the weight of the input data,the attention model is added to the decoder module to allocate the weight of the output data.This method improves the training effect of seq2 seq model,and solves the problem that seq2 seq cannot completely represent the whole sequence information.Finally,through the actual test of the system platform,the text content generated by word2 vec + LSTM + seq2 seq + attention training network proposed in this paper is effective and achieves the expected design goal.