How to Create an SRT File from a Video

The correct SRT workflow depends on where the words come from. Reuse an existing caption track when available, transcribe clear speech when no text track exists, and use a separate OCR workflow only when the words are permanently burned into the picture.

Published: August 9, 2026Last verified: August 9, 2026

Short answer

First identify the source. If the video has an accessible caption track, export or convert that track. If it only has speech, create timed text with transcription. If the visible words are burned into the image, use OCR. Then arrange the result as numbered SRT cues, review the timing, save as UTF-8, and test it with the final video.

For supported public video links, CueSift checks for a usable caption track first and can use speech transcription when eligible. It does not currently accept local files or read burned-in subtitles.

Try CueSift with a public link

Choose the workflow by text source

A selectable caption track exists

Extract or convert the track

Usually preserves the source timing and avoids generating new words from audio.

There is speech but no text track

Transcribe the audio

Create timed segments from speech, then proofread names, numbers, punctuation, and cue boundaries.

Words are part of the video image

Use video OCR

Sample frames, recognize text, remove duplicates, and rebuild timing. CueSift does not currently provide this workflow.

Original decision model based on source type. It is not a performance benchmark or a claim that every accessible video can be processed.

Create the SRT in six steps

Confirm access and rights

Make sure you can legally process and reuse the video. Public access does not automatically grant permission to republish its text.

Inspect the source

Look for a selectable caption track or embedded subtitle stream before spending time on transcription or OCR.

Retrieve or generate timed text

Extract the track, transcribe speech, or run OCR according to the source decision above. Preserve language and speaker information when available.

Edit cue boundaries

Split long passages at natural phrases. Each cue needs a start and end time and should not accidentally overlap the next cue.

Format and save

Number cues sequentially, use comma-separated milliseconds, keep blank lines, and save the file as UTF-8 with the .srt extension.

Play and validate

Test the result in the destination player. Check representative sections, the first and last cues, names, numbers, line breaks, and missing text.

A reusable three-cue SRT template

Replace the text and timing rather than copying these sample statements into a real project. The example demonstrates structure only.

1
00:00:00,000 --> 00:00:02,800
Add the first subtitle here.

2
00:00:03,000 --> 00:00:05,600
Keep the timing in chronological order.

3
00:00:05,800 --> 00:00:08,400
Save and test the completed file.

Frequently asked questions

Can I create an SRT without existing subtitles?

Yes, if the spoken audio is accessible and clear enough for transcription. The generated text still needs proofreading and timing review. Burned-in text requires OCR instead of audio transcription.

What information does an SRT file need?

Each cue needs a sequential number, a start and end timestamp separated by -->, one or more text lines, and a blank line before the next cue.

Can CueSift create an SRT from a local MP4?

Not currently. CueSift accepts supported public video links rather than local file uploads. Use a local media or transcription tool for files stored on your device.

How do I know the SRT is ready to publish?

Validate its structure, reopen it as UTF-8, play representative points in the final video, check names and numbers, and inspect the first and last cues. A file that parses is not automatically accurate.