Research Of Embedded Speech Recognition System Based On MiniGUI

Speech recognition is a hot research field of information technology. Its purpose is to realize the language interaction between man and machine to get rid of the shackles of the keyboard and mouse. There are several advantages. It will make intelligent machines for further development. It is no longer a dead machine. It has certain "personification". It can communicate with man; it makes the unnecessary input and output terminals"off", such as monitor, keyboard and mouse etc. At the present time, speech recognition technology has been used in telephone speech recognition systems, information search, and smart toys, smart home voice control system and automotive voice control device, and has achieved good results. Its development prospect is broad.Embedded speech recognition systems enable equipment to further miniaturization and covert of device. It meets requirements of future intelligent machines. Miniaturization and portability of performance will enable embedded system be applied to many other fields.Thesis aims to design an embedded speech recognition system, and to do some research about its promotion and application. The system uses OMAP5912 dual-core processor as the microprocessor; uses embedded Linux as the underlying operating system; uses MiniGUI as the interface; and uses the message queue as an inter-process communication to achieve unity with the speech recognition engine.Voice printer designed by thesis is to pass results of the speech recognition engine PocketSphinx in the embedded system MiniGUI and be displayed on the screen by the message queue, so that can apply audio recordings to meeting. Speech recognition used in the field of education will lead to the change of learning style. The voice learning system is aimed at children learning figure literacy and primary English pronunciation. Zodiac pictures will be loaded into the interface of MiniGUI to wait for the results of the match. If the match is consistent, pronunciation is correct, and then it will be into the next image;If the match is inconsistent, it indicates pronunciation is not correct, so it may require re-entry.Thesis makes some attempt on voice control, and designs interface of smart toys and smart home system voice control. By identifying the "LEFT" and "RIGHT" to control the picture of the car to move left or right 10 pixels, and then refresh and loading.