NoSightInbox: A Voice-Driven Email System with Pattern-Based Authentication for Visually Impaired Users

This paper presents NoSightInbox, a voice-first email management system designed to reduce reliance on visual interfaces and traditional keyboard input for visually impaired users. Built with the MERN stack (MongoDB, Express, React, Node.js), NoSightInbox integrates the Web Speech API for Speech-to-Text (STT) and Text-to-Speech (TTS) to enable end-to end voice interaction [4]. It introduces a novel click-pattern authentication mechanism—comprising left/right clicks, double clicks, and long-press gestures—as an accessible alternative to alphanumeric passwords, hashed with bcrypt and managed via JWT sessions [15][16]. The system features voice-guided navigation, speech-driven email composition, and audio feedback for all interactions, following WCAG 2.2 AA principles and ARIA best practices [3][8]. In a preliminary evaluation with visually impaired participants, NoSightInbox reduced task completion time compared to screen reader–based email workflows, achieved high voice recognition accuracy in quiet conditions, and was positively received on usability measures (SUS) and perceived workload (NASA-TLX) [21][22]. We discuss strengths, limitations (e.g., browser compatibility, noise sensitivity), and implications for accessible interface design. This work demonstrates the feasibility and benefits of a voice-first, pattern authenticated email client and outlines directions for enhancing robustness, personalization, and multilingual support.