<p>Audio Description (AD) is a vital tool for enhancing web accessibility, particularly for individuals who are visually impaired or blind. However, a significant challenge in the multilingual information retrieval domain is the real-time translation of alternative text (ALT) for images, which often fails to provide comprehensive descriptions. Existing systems typically offer brief ALT text, omitting critical visual details such as facial expressions, body language, and contextual backgrounds. This limitation hinders the ability of visually impaired users to fully understand and engage with visual content on multilingual websites. To address this issue, this paper proposes an Accessible Multilingual Audio-based Image Alternative (ALT) Description Model, designed to deliver real-time, detailed, and barrier-free image descriptions for visually impaired users. The models performance was validated through testing with 20 visually impaired participants, demonstrating its effectiveness in enhancing accessibility and inclusivity. Using the Jaro–Winkler string measuring algorithm, the similarity of ALT text across different language versions was evaluated. The analysis revealed that 75% of ALT texts remained unchanged during real-time web translation, highlighting the inadequacy of current systems in providing accurate and detailed descriptions. By addressing the limitations of existing approaches, this study contributes to the development of more accessible multilingual web environments, ensuring that individuals with disabilities can fully engage with digital content. The proposed model aligns with global accessibility standards and supports efforts to create a more inclusive digital landscape.</p>