Glossary

Caption terms, without the jargon

Short definitions for the words that show up in caption specs, client briefs and accessibility reviews.

Last updated: 31 October 2026

Auto captions: Captions generated automatically from a video’s speech by a speech-to-text model. They usually arrive with word-level timestamps and need a review pass.
Subtitles: Text that translates or transcribes dialogue for viewers. Subtitles can be a sidecar file (SRT/VTT) or burned into the video.
Closed captions: Captions for viewers who cannot hear the audio. They include non-speech information such as [music], [laughter] and speaker labels.
SDH: Subtitles for the Deaf and hard of hearing. SDH includes speaker identification, sound effects and meaningful non-dialogue audio.
SRT: SubRip Subtitle format. A plain-text cue list with an index, start and end timestamps, and one or more lines of text. Widely supported by YouTube, editors and players.
VTT: WebVTT, the web-native subtitle format. Supports cue settings, styling and regions; used by HTML5 video players.
TXT: A plain transcript without timestamps. Useful for show notes, descriptions and search indexing; not a subtitle format for playback.
Burned-in captions: Captions rendered into the video pixels during export. They appear on every platform and cannot be turned off. Also called hardcoded or open captions.
Soft subtitles: A separate subtitle track or file that the player can toggle. The video file itself does not contain the text.
Transcript: The text of the spoken audio, usually with timestamps. A transcript is not automatically formatted as readable caption lines.
CPS: Characters per second. A readability measure for captions; higher CPS means viewers have less time to read each line. Many caption guidelines target roughly 15–20 CPS for adult audiences.
WER: Word error rate. The percentage of words a transcription model gets wrong, including substitutions, deletions and insertions. Lower is better; audio quality changes the result.
Karaoke captions: Captions that highlight words or phrases as they are spoken, often with a color or background change.
Speaker diarization: The process of separating and labeling who spoke when. Useful for interviews, podcasts and multi-person videos.
Safe zone: The area of a vertical video where platform UI (captions, buttons, profile info) does not cover the picture. Keep captions inside it to avoid clipping.

Related reading