Title: Assessing Quality in Machine-generated Subtitles: A Case Study of “Netease Sight”

Abstract: Recent years have witnessed the rapid development of machine translation (MT) in China, which have brought significant changes to the subtitling industry, including the appearance of automatic machine-generated subtitles. This paper uses FAR model to assess the quality of a TED talk’s subtitles generated by Netease Sight, a MT platform, and compares them with human-translated subtitles. The study shows that the overall quality of machine-translated subtitles is much lower than that of those human-translated. Therefore, by maximizing advantages and minimizing disadvantages through technology development and post-editors, the future for machine-generated subtitles is still bright.

Keywords:: machine translation, subtitle, Netease Sight, FAR model, TED talk

Authors: Weiqing Xiao, Professor, School of English Studies, Shanghai International Studies University, Shanghai, China; Jiahui Gao, School of English Studies, Shanghai International Studies University, Shanghai, China.