Correct my Captions: An AI-Human Collaboration Workflow

Read full text Summarise with VIRAT
Authors: Derrel Fincher
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

As the April 24, 2026 ADA Title II deadline approaches, higher-education instructors are facing the task of improving captions on existing course videos. This poster presents a practical workflow to improve Canvas Studio captions in which AI performs most of the cross-checking between two independent transcripts: the original Studio captions and an AI-generated transcript from the extracted audio. The AI compares the transcripts and uses differences as a quality-check signal: segments with strong agreement remain unchanged, while mismatches, “inaudible/uncertain” markers, and glossary terms (names, acronyms, technical vocabulary) are automatically flagged for targeted human review. After review, the workflow outputs a revised caption file, optionally removing filler words, while preserving timing and updating caption text where warranted. The goal is a scalable, instructor-friendly approach that concentrates human effort where it adds the most value.

Fields of Study & Subjects

Subtitles and Audiovisual Media Artificial Intelligence in Education Translation Studies and Practices

Publication details

Journal / source
Pittsburg State University Digital Commons (Pittsburg State University)
Publication date
1 January 2026
Language
EN