Accessibility remains a critical challenge for visually impaired individuals, particularly in accessing textual information from their surroundings. Traditional Optical Character Recognition (OCR) systems often struggle with real-time performance and context-aware text processing, making it difficult for users to receive coherent and efficient auditory feedback. To address this, we propose a real-time text recognition and speech conversion system designed for visually impaired users. Our approach integrates a custom-trained YOLOCR model for text region detection, EasyOCR for optical character recognition, and Google Text-to-Speech for auditory output. Unlike traditional OCR systems that process text line by line, our method prioritizes text-dense regions, ensuring coherent and efficient information delivery. Experimental results demonstrate a detection accuracy of 75.2% mAp@0.5, an OCR Word Accuracy Rate of 92.5%, and a Text-to-Speech correctness of 95.0%, and a real-time performance of 18 FPS. This solution enhances accessibility by improving text recognition accuracy and reducing cognitive load for visually impaired individuals.