The current work aims to facilitate interaction with others to those with the inability to perform activities requiring motor skills or those who cannot speak. It proposes a modus operandi or a system based on Histogram of Oriented Gradients (HOG) and Support Vector Machine (SVM), which automatically identifies eye blinks in real-time to predict a lexicon. The system implements an auxiliary input that enables individuals to interact with others with the help of a device, where voluntary long blinks help in transition from a counter to a predictive table, while the short blinks are used to make the counter stop and select the lexicon. The system does not require prior manual initialization, special lighting, or previous face detection as it can calibrate it if the user is in the camera region and close. The proposed user interface makes the process of words detection by blinking easier with 74% accuracy.