Assessing the Effectiveness of Voice-Translation Tools for Translating Medical Information from English to Arabic

The study aims to assess the performance of Gemini, ChatGPT, and DeepSeek models in translating medical information from English to Arabic. Using a comparative descriptive design, the current study tests 40 English audio clips of medical information and evaluates their 120 Arabic translations using the Multidimensional Quality Metrics (MQM). The performance of each model was evaluated along two dimensions of translation quality: accuracy and fluency, including medical terminology, semantics, grammar, and clinical instructions. The comparative analysis revealed accuracy and fluency problems, including problems related to the accuracy of medical terms, omissions, clinical plausibility, and emergency communication. In the accuracy assessment, ChatGPT scored 18, Gemini 14, and DeepSeek 10 points. The current study stresses the importance of differentiating between linguistic fluency and medical accuracy and the significance of human supervision.