The Express Information Acquisition System Based On Speech Recognition

Under the background of "Internet +",the information system of China's express logistics industry has developed rapidly.However,in the process of collecting information in express delivery,manual entry is still the main method,which is inefficient and not intelligent enough.At present,the express company app mainly uses image recognition or specific text extraction technology to achieve information entry,and the accuracy and convenience of operation need to be improved.In order to improve such deficiencies,this thesis implements a express information collection system based on speech recognition.Based on the research of the theory and development status of express information system and speech recognition technology,this thesis takes intelligent express information collection as the research focus,and designs the system realization scheme and workflow based on speech recognition.The express information collection system consists of five modules: mobile acquisition module,identification module,data storage module,wireless transmission data module and transmission control processing module.The main work done in this thesis is as follows:First,analyze the needs of the express information collection system,and combine the relevant theoretical techniques to design the overall structure,hierarchical structure and logical structure of the system.Second,a self-built speech recognition system is used as an identification module.This thesis studies the language characteristics of express delivery information,focuses on improving the vector quantization process of self-organizing map,and designs and implements a speech recognition system based on improved self-organizing neural network.In this thesis,the corpus is first established,followed by speech preprocessing,feature extraction,codebook generation and template training.The speech recognition system before and after the improvement is compared and analyzed.The result shows that the highest recognition accuracy of the improved system is 93.3%.The improvement is obviously improved by 7.5%,the average recognition time of the sample is less than 0.16 seconds,and the relevant filesof the speech recognition system with the highest correct recognition rate are deployed in the remote server,which is convenient for the use of the mobile acquisition module.Next,the new mobile terminal application(WeChat applet)is used to realize voice collection,result display,information verification,uploading express information to database storage,and WeChat applet Bluetooth is used to transmit express delivery number to system hardware part.Next,study the Arduino core board circuit principle and control protocol,as well as the communication data frame format of Bluetooth and RFID.Design the Arduino main control flow,realize the Bluetooth hardware to receive the express number sent by Wechat applet Bluetooth,and then realize the deposit and reading of the express number in the express electronic tag,through the RFID data reading and writing steps such as query,anti-collision and sector authentication.Finally,through the function and performance test of the system,the test results prove that the express information collection system has flexible and intelligent information entry characteristics,and speech recognition response speed is up to 0.47 seconds.