Glossary

YouTube transcript glossary

Short definitions of the terms that come up when you compare YouTube transcript tools or hand a video to an AI assistant.

Updated 2026-11-05 · By the WithTranscript team · How we checked

Auto-generated captions vs speech-to-text transcripts: Auto-generated captions are the subtitles YouTube's own speech recognition adds to a video; a speech-to-text transcript is made separately by running the audio through another recognition model.
Speaker diarization: Speaker diarization is the step in transcription that works out "who spoke when": it splits the audio into turns and labels each turn with a speaker (Speaker A, Speaker B), without knowing their names.
SRT vs VTT vs Markdown transcripts: SRT and VTT are subtitle formats: numbered or timed cues that a video player shows on screen. A Markdown transcript is a document for reading: headings, paragraphs and optional timestamps, built for people and AI assistants rather than players.
YouTube chapters: YouTube chapters split a video into titled sections on the progress bar, either from timestamps the creator lists in the description or generated automatically by YouTube.
Transcripts and LLM context windows: A context window is how much text a language model can read at once, measured in tokens; a transcript has to fit in it (with room for your question and the answer) for ChatGPT or Claude to work with the whole video.
Word-level timestamps: Word-level timestamps give the start (and often end) time of every word in a transcript, rather than one time per caption line or paragraph.
Interactive (click-to-seek) transcript: An interactive transcript plays in sync with the video: the current line is highlighted as it is spoken, and clicking any line moves the video to that moment.
Transcript vs AI summary: A transcript is everything that was said, in order; an AI summary is a model's short rewrite of it. A summary is faster to read, but it can leave out or misstate details that only the transcript can confirm.
Closed captions vs subtitles: Closed captions transcribe speech and also describe other sounds (music, laughter, who is speaking) for viewers who cannot hear the audio; subtitles assume you can hear and only render the dialogue, often translated.
YouTube timestamp links: A YouTube timestamp link opens a video at a set moment by adding a start time to the URL, such as youtu.be/VIDEO_ID?t=90 for 1 minute 30 seconds.