Based On The Design Of Small-vocabulary Speech Recognition System And Speech Recognition

Speech recognition is an interdisciplinary science, and it is gradually becoming the key technology to man-machine interface technology in the field of information technology. The combination of the speech recognition technology and the speech synthesis technology makes it possible to operate with voice commands without using keyboard. The application of speech technology has become a competitive emerging high-tech industry.On the research basis of network robot projects, the paper tries to make some beneficial attempts on the speech signal processing technology, discusses the application of speech processing technology in the projects of network robots, and makes it the future technical reserves.The outline of the contents are as follows :1. Describes some basic theories of speech signal processing, introduces relative knowledge of speech synthesis, and researches the realization mechanism of speech recognition technology systematicly.2. Demonstrates the whole process of the speech recognition system from the pretreatment to Endpoint Detection, to feature extraction, and finally to the pattern matching.3. Meanwhile gives a detailed description of the whole process of the design of a speech recognition system based on speaker-independent and isolsted-word in small vocabulary, analyzes some key technologies of the system realization(choosing methods of endpoint detection, algorithm choice in feature extraction, the mode selection in pattern matching) and the related theories.4. Focuses on the feasibility of pattern matching algorithms in the embedded systems.5. Finally, shows the reasonable improvement achieved by the adoption of the improved DTW pattern in the design of mode matching in a speech recognition system based on speaker-independent and isolsted-word in small vocabulary with simulation results.