To promote the readability of multimedia and captions shown in mobile devices, we developed a system for processing listening materials to reduce and hide easy words which are the top 2200 highest frequent usage words, for better English proficiency and learners who have mobile phones. On the basis of information processing theory, this study set different thresholds for the most number of words displayed each time for diverse mobile devices. We further evaluated whether the results of caption filtering satisfies the thresholds. Results show that more than 71% of all the captions are hidden. Moreover, the proposed system achieves the expected values of the requirements for different device screen sizes, and readability improvement is remarkable.