This study presents a novel assistive system for individuals with visual impairments, designed to enhance independent mobility by describing the surrounding environment. The system integrates real time voice guidance, stereo vision-based distance estimation and object recognition. Objects are localized using a 3 × 3 grid camera segmentation. The distance estimation is done through a calibrated stereo vision setup. YOLOv8 is used for fast and accurate object detection and Vosk is incorporated into offline voice recognition to interpret user commands. Depth estimation was achieved using two iPhone 7 cameras, with a fixed baseline, and a polynomial regression model correlating pixel disparity with real-world distances. The system demonstrated a root mean square error of 122 mm, with 85% of estimations falling within a 5% tolerance margin. On the experimental hardware, the system achieved an average response time of 368.5 ms, indicating further optimization is needed for real-time use. While indoor testing yielded reliable results, outdoor performance was affected by dataset limitations. The results suggest significant potential for real-world deployment in wearable devices, contributing to independence for the visually impaired.