The accurate identification of medication is a cornerstone of personal autonomy and safety. However, for individuals with visual impairments, this task represents a significant challenge. Reliance on others or the absence of accessible labelling, such as the Braille system, not only compromises user independence but also significantly elevates the risk of preventable medication errors. Although assistive technology solutions exist, they are often not specialized for pharmaceutical recognition, require an internet connection, or lack a fully auditory interaction, thereby limiting their efficacy. To address these issues, we present MedScan, an Android mobile application specifically designed to assist people with visual impairments. MedScan integrates on-device Artificial Intelligence (AI), specifically Google ML Kit for real-time Optical Character Recognition (OCR) and Text-to-Speech (TTS) synthesis. The application processes video frames from the camera to detect and read text aloud, providing immediate auditory feedback. To ensure reliability without an internet connection, the recognized information is validated against an onboard pharmaceutical database, enabling fully offline functionality. This paper presents the design and implementation of MedScan, a functional prototype that demonstrates the viability of a robust, low-cost, and fully offline mobile tool. Our work represents a significant step towards improving autonomy and reducing medication risks for the visually impaired community.