People use these three terms as synonyms constantly, including people who work with video. They describe genuinely different things, and the differences start to matter the moment you need a file rather than just some text.
The short version
| Transcript | Subtitles | Closed captions | |
|---|---|---|---|
| What it is | The text of what was said | Timed text lines for display | Timed text including non-speech audio |
| Assumes you can hear | N/A | Yes | No |
| Includes sound effects | No | No | Yes |
| Identifies speakers | Sometimes | Sometimes | Usually |
| Needs timing data | Not necessarily | Yes | Yes |
| Typical use | Reading, searching, repurposing | Translation for viewers who can hear | Accessibility and sound-off viewing |
Transcripts
A transcript is a written record of the speech in a recording. In its purest form it's just text — a document you can read start to finish without the video.
Transcripts often carry timestamps, which is what makes click-to-seek possible, but the timing is a convenience rather than part of the definition. The point of a transcript is to be read, not displayed alongside the video.
Subtitles
Subtitles exist to solve a language problem. The assumption is that you can hear the audio perfectly well but don't understand the language — so subtitles render the speech, and only the speech, as timed on-screen text.
That assumption is why subtitles don't tell you a door slammed or ominous music started. You could hear that; you just couldn't understand the words.
Closed captions
Closed captions assume the opposite: you cannot hear the audio at all. They were designed for deaf and hard-of-hearing viewers, and that changes what has to be included.
- Non-speech audio.
[door slams],[ominous music],[laughter]— anything meaningful that isn't speech. - Speaker identification. Who is talking, which is essential when you can't distinguish voices.
- Manner of speech. Whispering, shouting, or speaking sarcastically, where it changes meaning.
"Closed" means they can be turned on and off, as opposed to open captions which are burned into the picture. In many countries captioning is a legal accessibility requirement for broadcast and commercial content.
The file formats
When people ask for "the subtitle file" they usually mean SRT or VTT. Both pair text with timings; they differ in age and capability.
SRT (SubRip) is the older and simpler of the two: numbered entries, a start and end time, and the text. It's near-universally supported, which is why it persists.
VTT (WebVTT) is the web standard, designed for HTML5 video. It handles positioning, styling, and metadata that SRT can't express.
Where YouTube fits
YouTube blurs all three. What it calls captions may be creator-uploaded with proper punctuation and speaker labels, or auto-generated with none of that. Its "Show transcript" panel presents timed caption data as something transcript-shaped.
Which one you're getting depends entirely on what the creator provided — the difference is covered in auto-generated vs. manual captions.