The Design And Implementation Of Automatic Language Recognition System

Language Identification is also known as Language Recognition. The pur-pose of Language Identification is to identify a specific language based on the input voice. Language Identification is widely applied in many fields such as network security, authentication and multilingual information service. Nowadays, with the increasingly open international environment and rapid development of network, Language Identification is facing challenges. The paper based on the previous research, start from the actual project, designs and implements a Lan-guage Identification System based on GMM-UBM(Gaussian Mixture Model-Uniform Background Model), and tests the performance both in NIST(National Institute of Standards and Technology)corpus and network environment. Starting with the system performance, The paper construct the GMM/SVM(Gaussian Mixture Model/Support Vector Machine) language identification system.The mainly work is as follows:⑴Designs and implements the language identification system based on GMM-UBM, and do some experiments. Analyze the performance of this system in different parameters of GMM-UBM models and duration of tested speech by experiments. And give the test result of on-line environment.⑵To improve the system performance, designs and implements the lan-guage identification system based on GMM/SVM, compares and analyses the performance both on GMM-UBM system and GMM/SVM system. In the last place, the fusion method of the two system is proposed in this paper.⑶This paper always consider the model computation, storage, and recog-nition rate to balance the overall performance. In the feature extraction, methods of computational preprocessing and rapid calculation are adopted. In the GMM model training and testing phase, Top N fast algorithm is adopted. These me-thods improve the speed and ensure the system's real-time requirements.