Research Of Traffic Information Detection Technology Based On Voice Recognition

With the development of the transportation industry and the prosperity of economy inour country, the traditional highway management mode has failed to keep pace with thedevelopment of the highway. At present, the poor traffic information detection is theimportant cause leading to traffic congestion and frequent accidents. How to realize theintelligent, informationize and efficient, becomes a problem to be solved now.The traffic information detection technology based on voice recognition, whichintegrated applied of digital signal processing technology, machinery and electronicsengineering technology, computer technology and so on, could make up for the traditionalvehicle detection equipment, such as electromagnetic induction coil detector, and videovehicle detector to a large extent. Traditional vehicle detection equipment is easy to damaged,tearing up the road, obvious effected by environment, and expensive. The new method couldgreatly improving the effect of traffic information detection, has a huge practical significance.The current audio signal recognition technology limited to technical issues, did not achievesatisfactory degree. Therefore, the traffic information detection technology based on voicerecognition has much space to improve.This topic detailed analysis of the mechanism and the characteristics of the vehicle soundsignal, on the basis of referring to the related literature at home and abroad. Comparing withthe speech signal, analyzed the feasibility of using the same method of speech signalprocessing. The Paper studied the pretreatment of the vehicle audio signal and itstime-frequency domain characteristic parameters. In the process of recognition, the Paperselected the Hidden Markov Model (HMM) as the basis of the recognition system, andexpounded the basic principle of HMM and its three basic problems with solution algorithms.And then put forward the theory scheme. Because of the limitation of experimental conditions,we chose the electret microphone and AD7606data acquisition module in the process ofvehicle detection system experiment, and collected a series of voice made by footstep andSagitar1.4t car. Extracted the signal feature by Matlab, then compared the two kinds of voicesignal to verify the feasibility of traffic information detection based on voice recognitiontechnology.