LARA: A Multilingual Voice-Activated AI Assistant System for Visually Impaired Users

Oice assistants are increasingly redefining the way humans interact with computers; however, many existing desktop-based solutions for visually impaired users remain limited by the lack of support for local and regional languages, thereby narrowing their accessibility. To address this gap, this paper introduces LARA, an AI-driven voice assistant developed using Python that enables users to perform a diverse set of tasks through simple voice commands. Designed specifically for visually impaired individuals and non-technical users, LARA offers an intuitive, hands-free interaction experience by combining speech recognition, natural language processing, and machine learning techniques. The system facilitates multilingual communication, allowing users to interact in their preferred language, and is capable of executing various system-level operations such as application control and information retrieval. In addition, LARA incorporates emergency assistance features to enhance user safety and independence. The development process followed a structured methodology that included requirement analysis, intent identification and modeling, speech-to-text and text-to-speech processing, seamless system integration, and accessible user interface design. Overall, LARA demonstrates the potential of intelligent voice-based systems to improve digital inclusivity. Future enhancements will focus on integrating robust offline speech and language models to ensure reliable performance even in environments with limited or no internet connectivity.