Voice-based Robot's Control Application Research

Since the industrial robot has been put into applications, it has played an important role in many fields such as social production and manufacturing. Currently, all kinds of robots which provide entertainment, teaching, and service are stepping into our daily life, they will also bring many convenience and pleasure to us. Voice has always be the most usual and natural ways of exchanging information in our daily life. It is essential to achieve voice-communication between people and robots, studies of voice-controlling have great practical significance and good application prospects.This paper mainly studies robot's voice-controlling both in speaker of dependent and independent. Firstly, the speech recognition technology of classical dynamic time warping (DTW) has been studied deeply in the way of speaker dependent. On one hand, the algorithm of DTW involves a great deal of computation in the process of speech recognition which will affect response speed of the system; On the other hand, the method of endpoint detection can't locate the endpoint of the speech signal precisely. If the DTW align the endpoint to match, it will bring some recognition errors. To solve this problem, so this paper proposes an improved DTW algorithm, which combines width restriction searching and relaxation of endpoint restriction is proposed in this paper, the experiments'results validate the improved DTW algorithm is much better than ever both in recognition rate and recognition speed.Secondly, the hidden Markov model which is the most popular speech recognition technology has been studied in the way of speaker indendent. Then, the production bases, ideas, algorithm of HMM and corresponding solutions has been analyzed deeply. Speech model which is produced in the process of speech training decided the recognition performance of the system on a great extent. So the method which combined the technology of smoothing model parameters and method of maximum information parameters estimation is proposed to improve the speech training process. The experiment result indicate that the improved training method can obtain better speech recognition effect.Finally, the designing process of humanoid robot motion is described. Applying the wireless Bluetooth technology to realize the communication between computer and humanoid robot motion, which make the motion controlling of robot more flexible. Based on the above work, combine the programming on VC++6.and Matlab to achieve humanoid robot's speaker dependent speech controlling. And realize speaker independent speech controlling to humanoid robot, using HTK software on the platform of VC++6.0. the testing results indicate that both of them reach the ideal speech controlling effects to humanoid robot.