Individuals with severe motor and speech disabilities often face significant barriers to engage in communication. The purpose of this study is to propose a non-invasive Human–Machine Interface that enables such individuals to convey messages using electrooculography signals. The system is composed of three main components: an eye gesture detection method, a customizable virtual keyboard, and a voice assistance module. The eye gesture module recognizes five distinct actions: four directional eye movements (right, left, up, and down) and voluntary blinks. These gestures are captured through surface electrodes and processed using signal filtering and thresholding techniques. A calibration test was designed to record individual eye dynamics and generate personalized control parameters for accurate gesture detection and interaction. The virtual keyboard includes adaptive features such as text prediction and completion, implemented through Natural Language Processing to optimize communication speed. Keyboard navigation is controlled with eye movements and letter selection is confirmed via double blinking, and finalized messages are converted into speech through an integrated text-to-speech engine within the voice assistance module. This project contributes to assistive neurotechnology by presenting a scalable, user-centered electrooculography-based communication solution for individuals with severe motor and speech impairments.