System-level Modeling and Performance Evaluation of Speech Recognition System Based on SystemC

With the development of IC technology, the system which increases complexity requires design at a higher level of abstraction.In this paper, systemlevel design flow and modeling methods based on SystemC for complex embedded systems are introduced.The method includes the hierarchical modeling methods, object-oriented modeling technology, and the "Top-Down" modeling methods, and provides system-level modeling guidance for designers.By using the system-level modeling method, the speech recognition system has been modeled successfully, and the performance of speech recognition system has been analyzed.The performance analysis results provide an effective reference on refined the speech recognition algorithm in further.