Live streaming captions and subtitles and caption-quality standards were among the key topics discussed at a recent meeting of nearly three dozen experts from the broadcasting, captioning, and technology sectors.
CCSUBs featured participants from the media supply chain, representing content platforms, service providers, and viewers. The group’s goal was to share insights on some of the technical challenges facing the captioning and communication access industry and how the community can help address them. Discussion points featured topics suggested by VITAC, Netflix, Venera, Imagica, Gallaudet University, and others, including interoperability concerns regarding captions and subtitles in streaming as well as those around measuring caption quality.
VITAC was among the sponsors of the event, which was held during last month’s National Association of Broadcasters (NAB) show in Las Vegas. Below are a few of the items that were up for discussion.
Live streaming captions and subtitles
Though streaming pre-recorded content has been around for more than a decade, the CCSUBs group noted that streaming live events across multiple geographies is relatively new. As such, several challenges remain, including:
- Making sure captions and subtitles are in sync with the audio and video
- Avoiding conflicts between the caption and subtitles placement and onscreen graphics, like scores, stats, or other scrolling information
- Supporting multiple simultaneous languages, including non-European languages (the group cited Grand Prix Motorcycle Racing events are sometimes broadcast live in dozens of languages)
Quality Control
The group discussed the challenges with measuring the quality of captions and subtitles and noted, for example, that “word error rate,” one of the common methods of measuring caption accuracy, is not a wholly sufficient metric; that the definition of quality can differ between viewers; and that automated text-to-speech remains relatively inaccurate in languages like Japanese where homonyms are common and text can be displayed horizontally or vertically. Identifying a standard for quality metrics was recognized as a topic of further discussion.
Interoperability testing and best practices
The group discussed interoperability between media players – how captions and subtitles are viewed within various streaming platform apps as well as across different streaming platforms – and the limitations of end-to-end capabilities, with subtitles and captions often being displayed onscreen, bottom-centered, in white text even though media players are capable of displaying different fonts and styles.
The group noted that interoperability testing and further discussions of best practices would be needed to improve live captions and subtitles across the streaming ecosystem. The topic is slated to be discussed at future CCSUBs meetings.
Viewer feedback needed
The need for viewer feedback continues to be necessary to improve captions and subtitles since, as the group noted, “the litmus test for timed text is a satisfied user.” Viewer feedback is also critical because different users have different preferences, not only in terms of caption style but also in whether they prefer verbatim or condensed text. Allowing caption and subtitles customization, the group noted, is an essential aspect of the viewer experience.
What’s next?
CCSUBs will hold its next meeting on interoperability challenges in live captioning and subtitling for streaming on June 10 in Berlin, Germany, during the Media Web Symposium. The event will recap the discussions and conclusions of the NAB 2024 meet-up and gather feedback from the participants on the topics of live subtitling and captioning as well as on quality measurement of subtitles.