Development Of Mongolian Voice Interaction System For Intelligent Robots

The voice interaction system is one of the most popular research hotspots in recent years.With the rapid development of living standards and science and technology,intelligent robots have gradually entered the lives of the majority of Mongolian people.They hope to communicate with intelligent robots through Mongolian.In response to this demand,this paper develops a Mongolian voice interaction system for intelligent robots.The voice interaction system is mainly composed of three modules: speech recognition,question answering system and speech synthesis.Mongolian speech recognition and Mongolian speech synthesis have already matured products.However,the research of the Mongolian question answering system is still in its infancy.First of all,this paper builds a Mongolian question-answer corpora and constructs a question answering model using LSTM-LSTM,LSTM+Attention-LSTM,BiLSTM+Attention-LSTM and BiGRU+Attention-LSTM respectively.The experimental results show that the question answering model with BiLSTM+Attention-LSTM has the best performance and uses this model to build a Mongolian question answering system.Secondly,this paper builds a Mongolian voice interaction system based on Android system.The Mongolian question answering system was deployed to the server,and the Mongolian voice interaction system for intelligent robots was completed through the coupling with Mongolian speech recognition and Mongolian speech synthesis.The system test results show that the system achieves better performance.For the Mongolian speech problem input by the user,the system can feedback the answer to the user in the form of Mongolian speech,and the answer is highly correlated with the question.