The accuracy of automatic and human live captions in English
Bibliographic record
Abstract
Closed captions play a vital role in making live broadcasts accessible to many viewers. Traditionally, stenographers and respeakers have been in charge of their production, but this scenario is changing due to the steady improvements that automatic speech recognition has undergone in recent years. This technology is being used to create intralingual live captions without human assistance and broadcasters have begun to explore its use. As a result, human and automatic captions co-exist now on television and, while some research has focused on the accuracy of human live captions, comprehensive assessments of the accuracy and quality of automatic captions are still needed. This article airs this matter by presenting the main findings of the largest study conducted to date to explore the accuracy of automatic live captions. Through four case studies that included approximately 17,000 live captions analysed with the NER model from 2018 to 2022 in the United Kingdom, the United States, and Canada, this article tracks the recent developments with unedited automatic captions, compares their accuracy to that achieved by human beings, and concludes with a brief discussion of what the future of live captioning looks like for both human and automatic captions.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.008 | 0.057 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.003 | 0.003 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.005 | 0.005 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.005 | 0.003 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".