Arm-based Platform For Engine Voice Recognition Technology And Research To Achieve

With the rapid development of super highway and automotive engineering, the automobile flow becomes bigger and bigger, more and more large-scale rechargeable parking lots and highway toll stations appear, both of them need an effective intelligent management system, which can judge different types of automobiles exactly. Taking advantage of Sound detecting method, we can distinguish different types of automobiles from a lot of automobiles which are driven on the road effectively. This method can be applied for large-scale rechargeable parking lots and highway toll stations to charge different types of automobiles exactly. At present, the automated fare collection system has not been used practically. With the promotion of the demand of automation, the automated fare collection system will be come true and used widely.This work is base on simulation research the choice of feature parameters and identifying model. This paper select Samsung S3C2410as the automobile engine sounds recognition system’s microprocessor and Linux as the automobile engine sounds recognition and test embedded operating system.This paper summarizes the basic theory of automobile engine sounds signal processing. This paper select the parameters of the automobile engine sounds through simulation analysis the sample frequency, the frame length, feature parameters and recognition network’s parameters. The hardware application programming depends on the simulation analysis. This paper summarize hardware memory and JTAG port and made a detailed instruction in Boot loader、file system and Linux kernel. I write the automobile engine sounds recognition program by C language. This program included MFCC extraction function and recognition function. I consider the requirement of real-time and the limit of system resource. The programming is directly readied in by text for accelerate recognition rate. This paper build the cross compiling environment after testing the C program and add the recognition program into file system and compile audio driver into kernel[1].This paper completed the boot loader porting, kernel porting and root file system porting. I completed the automobile engine sounds recognition system and real-time testing.The use of embedded system as the engine sound recognition platform can fit for different places. The paper completed the recognition of broad categories automobile. Future missions is more fine classification the engines and increasing recognition rate. Another mission is improving the robustness to fit for the noise environment.