Research And Development Of Blind Speech Recognition System Based On Leap Motion

In our life,there is such a group--the deaf and the dumb.According to relevant institutions of China's deaf population about twenty million,in the vast crowd,blind language became one of the main methods of communication between them.Because of its own particularity and low popularity,it is a barrier to communication between the deaf and the deaf.This article on how to solve the problem of communication between the deaf and the deaf and blind identification system,is developed by the deaf blind real-time translated into language can let the deaf and dumb people understand the phonetic form.First of all,we through the body feeling equipment Leap Motion will sign the hand movements of acquisition,combination of using Unity3D software for data processing and storage dynamic hand gestures related data information.Then,we will with the dynamic signal data stored a 3 d coordinate conversion processing complanation,gestures in 3 d space coordinate information can be converted to coordinate information in the two-dimensional plane,this is done to reduce the number of data processing,which can speed up our system identification.Finally,we will have the complanation dynamic gesture data to a gauge,a formula according to relevant rules of the data information in the form of structured as an integer,the aim is to decrease the rate of some fuzzy gestures deterrent to,and also for using the HMM model for identification provides a premise condition.This article also USES the HMM model to identify dynamic gestures by looking at the relevant literature.We designed a set of blind and real-time recognition tests based on the HMM model.Through the analysis of experimental results,we proposed a two-channel model based on the HMM model,the coordinate feature model and the pen drawing feature model.In this paper,we have carried out the experimental design of the two feature models,and analyzed the advantages and disadvantages of the models under various models based on the experimental results.We also hope to find a model that will improve the recognition speed and accuracy of the recognition system by further research.Because of the differences between users and gesture recognition for different people,it is a standard hand movements will be all kinds of differences,this also increases the recognition accuracy.In the process of acquisition signal data to produce a large amount of data,due to the large amount of data on the requirement of the algorithm has the very big enhancement,we hope in the future research and development for the recognition speed and accuracy in the process of the above,we are going to have a leap,we hope to find a new technical way to fundamentally solve this problem.