Design And Implementation Of Language Recognition System Based On I-vector

In recent years,with the rapid development of information technology and the continuous deepening of international communication,the demand for multilingual communication is increasing,and the importance of language recognition technology in the field of intelligent speech is also increasing rapidly.However,the current language recognition technology mainly stays at the algorithm research stage,existing language recognition systems also mainly relies on audio files uploaded by user for recognition,and the interaction method is relatively simple and cumbersome.If a convenient and efficient language recognition system can be realized,the language recognition technology can be one step closer to application.This thesis aims at the problem that the current language recognition system is less implemented and the existing system interaction mode is simple and cumbersome,an i-vector-based language recognition system is designed and implemented.The specific research content is as follows.This thesis has completed the design of language recognition system based on i-vector.A hierarchical design scheme with high cohesion and low coupling is adopted,and the overall design and hierarchical design of the language recognition system are completed from the system requirements.The system includes five levels:data acquisition layer,data transmission layer,algorithm analysis layer,system management layer and system visualization layer.The data acquisition layer is composed of real-time data acquisition modules and offline file acquisition modules.The data transmission layer is composed of voice data transmission modules and system signaling transmission modules.The algorithm analysis layer includes voice preprocessing,feature extraction and fusion,i-vector extraction,and language discrimination.The system management layer includes three modules:user management,equipment management,and data management,and the system visualization layer includes two parts:front-end interactive page display and chart visualization.This thesis has completed the implementation of language recognition system based on i-vector.Under the condition of completing the design of the language recognition system,the implementation of the data acquisition layer,data transmission layer,algorithm analysis layer,system management layer and system visualization layer in the system has been completed successively.The data collection layer supports two voice data collection methods:microphone module pickup and offline files upload.The data transmission layer is responsible for transmitting voice data and system signaling to the service background,and then the service background calls the language recognition algorithm of the algorithm analysis layer to classify and recognize the voice data.The system management layer and system visualization layer support users and clients to access the system more friendly and manage language identification data.Finally,a detailed test and analysis was carried out,the results show that the system can complete the whole language recognition task in 14 seconds,and the recognition accuracy is more than 87%.