ScholarWorks@숙명여자대학교

초록

Video lectures, which account for a large part of the non-face-to-face lecture system, conduct classes based on visual data. However, blind students cannot find out the contents of the screen and cannot understand the class properly, and it leads to the infringement of their rights to study. For this reason, this paper proposes a service that uses the automatic voice commentary system of the online video lecture screen to enhance the understanding of the class of visually impaired students and ensure the right to learn. We proceeded with requirement analysis and a prototype test to confirm the validity of the proposed method. First of all, it's necessary to know the timing of the appearance of each slide in order to provide an appropriate description for each screen. Thus, we designed the similarity discrimination algorithm between the frame and the image to identify the timing of screen transition and the slide number. Then, the system creates commentary files based on text, images, and tables within each slide, and generates video files that read the commentary files by voice using Text-to-Speech technology. Finally, it merges the original lecture and commentary videos as one video according to the order of slide appearance. After conducting the usability evaluation, we confirmed that the understanding of the lecture content increased when students listened to the explanation on the screen with the proposed service. It was notably more effective when providing descriptions of objects such as pictures and tables. Through this, we found that the proposed system was helpful in learning for visually impaired students who took video lectures.

키워드