The Research Of Application Speech Recognition Technology In Feature Functional Test System

With the rapid development of technology, intelligent human and machine interfaces are used in appliances in people's normal life. Currently, a lot of appliances have the function that can give warning messages to the user in sound which improves the efficiency of human and machine interaction. The feature functional test is a key node of production and it is important to assure the high quality of the products. To test the prompt tone feature, a feature functional test system of prompt tone is exploited based on the speech recognition technology. Mainly works in this paper:Firstly, this paper introduces the basic theory of speech recognition and then analyses some feature parameters such as Linear Prediction Coefficients,Linear Prediction Cepstral Coefficients and Mel Frequency Cepstral Coefficients and chooses the MFCC used in this system.Secondly, this paper also makes a point research on model-matching based on DTW and VQ and proposes a new method of the optimal codebook searching of VQ based on multi-classic algorithm using support vector machine.Then this paper analyses the main requirements of this system, designs the hardware and software in detail and chooses the graphic programming language LabVIEW as the software development platform and then implements the feature functional test system with the platform. Followed, this paper realizes automatically speech signal acquisition, and uses multi-thread of LabVIEW to calculate the feature parameters in a parallel way and short 1/4 of the calculating time which improve the efficiency of the feature functional testing.At last, it tests the system and analyses the results and it shows that the system can meet the test requirements and system running stably.