Subtitles are a synchronized text track that displays a translation of the spoken dialogue for viewers who do not speak the original language; this is their defining distinction from captions, which render the same language as the audio and include non-speech sounds for accessibility. In practice the terms are often used interchangeably in informal contexts, but the technical and regulatory distinction matters: WCAG accessibility requirements apply to captions, not subtitles. In the e-learning localization pipeline, subtitles are generated by applying machine translation (MT) to a source-language ASR transcript, then time-aligning the translated segments to the original video timestamps. The file formats are the same — SRT, VTT, TTML — regardless of whether the track is a caption or subtitle. Subtitle quality depends on both ASR accuracy (errors in the source transcript propagate) and MT quality; post-editing by a human translator is recommended for certification-level or regulated content. A practical constraint is line length and reading speed: translated text is often longer than the source, requiring the subtitle segments to be re-timed or the line count to be managed so learners can read at normal speed. For multilingual catalogs, managing multiple subtitle tracks per video and keeping them in sync with course updates is a content-operations challenge that benefits from version control and automated tooling.
