Design Of Speech Recognition System On Vehicular Multimedia Platform

Nowadays, automobile's entertainment function more and more receives people's favor, but this actually greatly deepened the hidden dangers of traffic accidents, especially traffic safety is paying great attention and drive on the phone is a dangerous, moreover, operating a wide variety of entertainment. In order to protect the safety of drivers, improving the means of human-computer interaction is necessary. Speech recognition is the best solution to meet this requirement. Now, along with science's swift development, each kind of intelligent equipment have already moved toward the practical application, this is also the best carrier of vehicular speech recognition system's.This paper mainly researched how to construct speech recognition and control system on vehicular multimedia platform, provides the good human-computer interaction environment for the driver. This paper describes the basic principles of speech recognition and the classified of speech recognition technology. Come to vehicular multimedia speech recognition system is based on small vocabulary isolated word speech recognition technology, and particularly analyze the critical technique. Based on this, in view of the vehicular multimedia platforms, improved the endpoint detection algorithm; Compare the merits of LPCC and MFCC, consider the algorithm accuracy and computational, Selected the voice characteristic parameter which is suited in the embedded platform. This system used DTW to be the matching method of pattern recognition. After the experiment, above several algorithms are quite suitable for the embedded system. The next work is to design an ARM-based vehicular multimedia speech recognition system according to the proposed speech recognition algorithms.Finally, on the basis of summary of this paper, the existing problems of this design have been analyzed, and the direction of further researches has been pointed out.