Design And Implementation Of Conference Content Transcription System

Nowadays,countless conferences are held all the time in the world,ranging from the weekly paper study exchange meeting in a laboratory to the United Nations General Assembly,which concerns the survival and development of hundreds of millions of people.Every meeting held has its special meaning,and recording the content of the meeting in a form that is convenient for review means to a certain extent the preservation of important meeting results.The traditional paper-pen method is used for recording.Because the speed of writing is generally slower than the speed of speech,under normal circumstances,only the main points can be recorded and cannot be recorded completely.In important occasions,it can be seen that professionally trained typists use professional typing equipment to record.The disadvantage is that the cost of training professional typists is very high.The disadvantage of recording the meeting with the help of recording equipment such as a voice recorder is that the recording in sound format is not easy to review compared with the text format;if dictation is performed afterwards,it also requires a lot of manpower.With the development of machine learning and deep learning technology in computer technology,speech recognition technology and speaker recognition technology are gradually or have been applied in specific fields.These two technologies are innovatively applied to the meeting content transcription scene.As long as there is any recording device to record the meeting content,the meeting content can be finally converted from the recording into the text form of who said what,when.In this way,the content of the meeting can be fully recorded,which is convenient for review and frees up manpower.This paper designs and implements a conference content transcription system.Including research on the implementation of speech recognition technology and speaker recognition technology,improve the speaker recognition technology that is not yet mature,improve the accuracy of recognition,write a client with a visual graphical interface,and a server that integrates speech recognition and speaker recognition functions,And carry out network communication through network protocol.