Design And Implementation Of Speech Interaction System For Educational Robot

Speech is the most commonly used way of exchanging information in human daily life.In recent years,with the rapid development of speech technology,speech interaction technology has been widely applied in human learning,life and work,and the terminal of speech interaction is no longer limited to smart phones,computers and other devices with the advances of robot technology.It has become one of the development trends in the field of human-computer interaction to increase the function of speech interaction to the robot and make the robot have a more natural and friendly human-computer interaction mode.The research work of this paper is based on the embedded platform of the educational robot motherboard raspberry pie,because of the limited computing resources and processing ability of embedded platform,under this requirement,how to build a practical speech interaction system for educational robot,which can be applied in specific teaching scenarios,is the main research content of this paper.First of all,this paper briefly introduces the current domestic and foreign research status of speech interaction from two aspects of technology and application,and elaborates the related concepts and common methods in the speech interaction technology,which lays the foundation for the following research on the speech interaction system for educational robot.Secondly,the function module of speech interaction model for educational robot is designed and implemented.After comparing and analyzing the characteristics of different open source speech tools and frameworks,the speech synthesis tool Ekho and the speech recognition tool Sphinx,which are suitable for educational robots,are selected to deploy the speech interaction environment,and implements the speech interaction function through installation,configuration and Linux C programming.In the process of speech synthesis,the synthesis of characteristic speech is realized by replacing the original pronunciation database.In the process of speech recognition,according to the speech recognition and interaction requirements of professional vocabularies in the teaching of specific subjects,the method of training language model and establishing phonetic dictionary based on small and medium vocabulary is studied.In addition,in order to understand the influence of different configuration parameters on the training of acoustic mode,this paper compares and analyzes the effects of parameters such as Gaussian mixture number,HMM state number and other parameters on the recognition results through experiments,and selects the relatively superior parameters according to the experimental results,and it is also proved that choosing the appropriate parameters is very crucial to the training of the model.Finally,this paper selects the teaching scenarios of question answering with ancient poetry and four arithmetic operations to test on the platform of educational robot.Through the use and test of different users in the scene of ancient poetry,the recognition accuracy rate is above 86%,the test results show that the speech interaction system can basically meet the requirements of the practical application of speech interaction oriented to a specific subject,and allows students to interact with robot without network,but there is still a better promotion space in the performance and function design of system.