Visually Impaired Voting Aids using Speech Processing and face Recognttion

Karthikeyan, M. Mahendran, S. A. K. Jainulabudeen and K. Somasundaram

Department of CSE, Panimalar Engineering College, Chennai.

Abstract: In proposed method, first voice is given as the input (The voice will be the adhar card number) and converted in to text in order to compare with the database. If the adhar card number matches, then the face recognition will be taken and compared. If both adhar card number and face will be authenticated, then the already recorded voice will be played. The recorded voice will have the details about the political parties, and according to the recorded voice we want to give the input. Then the vote will be done accordingly. In this system HMM, GMM and SOM (self organized mapping) filters are used for speech. Segmentation analyzes the system MFCC with SVM trained samples of speech and recognized text rate results given via MATLAB IDE. Efficient results give which compare to existing HMM model. For face identification using viola Jones algorithm is used.

Keywords: Mel-Frequency cepstral coefficients (MFCC); Hidden Markov Model (HMM); Gausian Mixture Model (GMM); Acoustic vectors and models; Dynamic Time Warping

[ HTML Full Text]

Back to TOC