Design of language model of speech recognition based on HTK and its performance analysis

In the Large Vocabulary Continuous Speech Recognition(LVCSR) the statistical language model plays an important role in the entire speech recognition system.The principle and related techniques of the statistical language model are analyzed and the model suitable for the large and continuous speech recognition system is designed by using the HTK.The complexity and accurate recognition rate of the model are analyzed.The related factors affecting the performance of the model are pointed out.