A Study Of The Prosthesis Calibration Based On Speech Recognition And Electromyographic Signal

The number of amputees is increasing each day.The research of the prosthesis has become a hot research both at home and abroad. Developing an autonomy and flexibility prosthesis for amputees has a great meaning.EMG signals that can be used to control the prosthesis is limited.For this issue, the paper proposes an idea that using voice signal combines with EMG signal to control prosthesis.The main contents are as follows:(1) Presenting a detailed description of the voice signal and EMG signal integration proposal. Using the voice signal of amputees as prosthetic auxiliary control signal combines with the EMG by generated from amputees residues. To some extent, the prosthetic calibration mode control can be more freedom and flexibility.(2) Voice signal recognition analysis module: the basic principle of voice signal,MFCC extracting characteristic parameters, HMM training and recognition are introduced in detail. And this paper experiments the four actions needed voice instructions in HTK environment,so as to verify the feasibility of using HMM algorithm for voice recognition system design,and analyze the corresponding data.(3) EMG signal recognition analysis module: the contents of EMG signal acquisition, classifier algorithm are introduced in detail. Aiming to adapt to the time-varying characteristics of EMG signals in pattern classification, a novel method: a self-enhancing classifier is improved.This adaptive method for dynamic updating the parameters: covariance matrix and mean vector is based on the traditional classifier(LDA and QDA).And this method is used to update parameters during testing time. Experiments show that this method reduces the computing complexity, and improve the classification performance.