SubRip subtitle file (SRT)

A SubRip subtitle file, usually called an SRT file, is a plain-text file that stores captions for a video. The name comes from the SubRip program, and the extension is .srt. Most subtitles distributed on the internet use this format ¹. It is simple enough to open in any text editor, which is one reason editing tools and video platforms accept it.

Each caption is a numbered block

An SRT file is a list of blocks, and each block has three parts. The first line is a sequence number. Numbering runs sequentially, starting at 1 ². The second line gives the start and end times of the caption, joined by an arrow. The following lines hold the caption text, and a blank line ends the block. Because the format has no single formal standard, tools differ on small details, but this three-part block is the common core.

1
00:00:01,000 --> 00:00:03,500
Dictation writes as you speak.

2
00:00:04,000 --> 00:00:06,000
Check the draft before you send it.

Timecodes use a comma for milliseconds

The times in an SRT file read hours, minutes, seconds and milliseconds. The separator before the milliseconds is a comma, not a period ³. That detail is the most common cause of a failed import, since a tool that expects a period will reject the file or misread the times. The W3C based its WebVTT text track format on the SRT file format, so the two are close relatives ⁴, but WebVTT uses a period.

A concrete example

Suppose a transcript of a meeting has a sentence that spans four seconds. A transcription tool writes that sentence as one block, with a start time and an end time. A video editor reads the block and shows the text on screen for those four seconds. If the times are off, the caption appears early or late, which is why the times usually come from forced alignment or from a careful timing pass.

Why it matters for dictation

Dictated text is not a subtitle until someone adds times and breaks it into blocks. If you dictate a video script or a talk, the words need to be split into short caption lines and timed. Many speech tools can export a transcript with timestamps, and an SRT export is then a formatting step. Keep each caption short enough to read in its time, because a long caption on a short block cannot be read in time. For how captions differ from subtitles and what the common rules are, see our guide to captions and subtitles.

SRT files carry the words of a clean verbatim transcript most often, because captions are read, not heard. Their times rely on timestamps that match the audio, and an error in the text, measured by word error rate (WER), shows up on screen as a wrong word.

Related terms

• Timestamp (transcription): the times that each block uses.
• Forced alignment: the method that times a known transcript.
• Closed captions (CC): captions that viewers can turn on and off.

Guides that cover it

• What is an SRT file?
• Captions and subtitles: formats, rules and tools
• Video to text: transcribe videos and YouTube