This study explored the role of live transcripts in online synchronous academic English classrooms by focusing on how automatically generated live transcripts influence the learning outcomes of lower-proficiency and higher-proficiency learners and on their perceptions towards live transcripts. The study adopted a 2 × 2 factorial design, with the two factors being learner proficiency (high vs. low) and availability of live transcription (presence and absence). The participants were 129 second-year Japanese university students from four synchronous classes taught on Zoom by the same teacher under an academic English reading course. Learning outcomes in this study were evaluated according to the course syllabus through grades and participation in class activities. A questionnaire consisting of nine Likert-scale questions and a comment box was administered to explore participants' perceived usefulness of, perceived ease of use of, and perceived reliance on live transcripts. Results showed that contrary to previous studies reporting the effectiveness of captioned audiovisual materials in L2 learning, live transcripts as a special type of captions were not effective in promoting the grades of learners of either proficiency. However, it significantly improved the activity participation of lower-proficiency learners, but not that of higher-proficiency learners. Questionnaire results showed that there were no significant differences between learners of two proficiencies in their perceptions towards live transcription, which contradicts previous findings that lower-proficiency learners tend to rely more on captions. Besides enhancement of lecture comprehension, participants reported innovative uses of live transcripts such as screenshots with transcripts for notetaking purposes and transcripts downloaded for later review.