Research On Embedded Speech Recognition System Applied In Robot

Speech recognition technology is to transform the voice signal into corresponding text or control command through the recognition and understanding of robot. The speech recognition is a cross-discipline, and is gradually becoming the key-technology of human-machine interaction in information technology. The combination of speech recognition and speech synthesis enables man more convenient to operate the computer through speech order without the keyboard. It is the simple way to realize man-machine interaction through voice control, and also can make the demands for the level of the operator's professional technologies greatly reduced lowering. Moreover, the communication is essential between operator and robot to make robot have human intelligence truly. Therefore, there are practicality and foreground to be engaged in the research on speech control.The principal contents of this thesis are as follow:Firstly, this paper presents the fundamental knowledge of speech recognition and several speech recognition algorithms, include: speech signal pre-processing, noise suppression, endpoint detection, feature extraction, DTW and HMM recognition algorithm, etc..Secondly, According to the model structure of speech recognition system, the endpoint detection method is based on SAM-SAZR (Short-time Average Magnitude and Short-time Average Zero-crossing Rate). The system adopts Mel Frequency Cepstrum Coefficient (MFCC), which considers fully sense of hearing characteristic, as voice signal eigenvector. The wavelet packet adaptive threshold function is adopted to denoise the noised voice signal. Moreover, the discrete hidden markov model (DHMM) is adopted and forward-backward procedure, viterbi algorithm, and evolutionary Baum-Welch reestimation algorithm are utilized to train and recognize the speech signal.Thirdly, we give the hardware and software design of the system according to the characteristics of the embedded system and the particular requirements of the project. PXA270-Microprocessor produced in Intel is selected as the core chip in the embedded speech recognition and interaction hardware system. At last, we realize the voice control function of motion robot.Finally, experiments are done to validate the Embedded Speech Recognition System, to test the speech recognition rate and the recognition stability under different situation, then to test the capability of speech input and to have a speech control test for the motion robot. The feasibility and validity of improved recognition algorithm have been further confirmed.