Research And Implementation Of Speech Recognition System For Mobile Robot

With the development of society, the robots with many kinds of function are widely used in various production and manufacturing industries. Meanwhile, intelligent robots which possess lots of other application have emerged in our everyday life, as a result, the application of the robot gradually moves towards civilian. In the process of robots control and application, the traditional keyboard input method has been unable to meet the needs of people. Therefore, as the peculiar result of the development of human society, the language which is the most natural and most direct form of communication will become the mainstream way of the future human-machine communication. The research of speech recognition for robot control will play a key role in the development of intelligent robots. The issue of speech recognition system for mobile robot control in real-time is deeply studied in this paper.First, according to the process of the speech recognition, each part which can affect the speech recognition has been researched. The research consists of the following parts:1. The research on the pretreatment and endpoint detection part of the speech signal and the comparison of the effect using different parameters as the efficiency threshold in the double threshold endpoint detection algorithm.2. The research on the several feature extraction methods, such as LPCC and MFCC, etc., and the characteristics of each parameter have been compared by experiments.3. This paper studies the speech recognition algorithm, including dynamic time warping (DTW) algorithm and hidden Markov (HMM) model algorithm, the problems emerged in the use of each algorithm, and the improvement of the traditional method and the comparison of recognition results of each algorithm and its improved algorithm by experimental simulation.Second, this system is the speech recognition system for mobile robot control, and according to the characteristics of the system, the speech recognition system used by this system is a speaker dependent, isolated word and small vocabulary embedded speech recognition system. After the consideration of the requirements of this system and comparison of experimental results, the speech recognition algorithms which is suitable for this system has been selected, and the programming work has been completed using VC programming language.Finally, the speech recognition system for mobile robot control has been designed and realized. In order to guarantee the real-time performance of the system, the real-time automatic recording system has been designed based on VC programming language. After the connection and debugging of the system, this system has been tested in the laboratory environment and the experimental results have achieved our anticipated effects.