Your caption file is a first draft you've already written

If you’ve ever downloaded the captions for one of your own videos and opened the file, you’ve seen something like this:

47
00:03:12,480 --> 00:03:15,020
so the thing nobody tells you about

48
00:03:15,020 --> 00:03:18,340
sourdough is that the starter is the easy part

It’s a mess. It’s also, buried in there, every word you said in the video, in order, already typed out. For a ten-minute video that’s somewhere around 1,500 words. That’s a blog post. Or three newsletters. Or the show notes you keep meaning to write.

Most creators never use it, because getting from that file to readable prose feels like a chore. It’s less of one than you’d think.

What’s actually in an SRT file

An SRT (SubRip) file is a list of numbered “cues”. Each cue has three parts:

1. A sequence number: 47.
2. A timecode showing when the line appears and disappears: 00:03:12,480 --> 00:03:15,020. Hours, minutes, seconds, and milliseconds after a comma.
3. One or two lines of text.

Cues are separated by a blank line. That’s the whole format.

VTT (WebVTT) files are close cousins. They start with a WEBVTT header, use a period instead of a comma before the milliseconds, and sometimes include styling tags or speaker labels. Same idea underneath.

To get readable text out of either, you remove the numbers, the timecodes, and the blank lines, then stitch the text back together. Doing that by hand in a long file is tedious and error-prone. Doing it with a regex in a text editor works, until a file has a slightly different format and you’re debugging a regex at midnight. A cleaner tool does it in one go.

Getting a clean transcript

You can grab captions from YouTube Studio under Subtitles for any video that has them. If your auto-captions are rough, it’s worth fixing names and jargon in Studio first; you’d want those corrections in the transcript anyway.

If you use Whisper or another transcription tool locally, it can also output SRT or VTT directly.

Then strip it down. Drop the file into the SRT Cleaner, pick plain text, and you’ll get the words without the scaffolding. It runs in the browser, so the transcript of your unreleased video doesn’t get uploaded anywhere along the way.

Spoken words are not written words

Here’s where most people get disappointed. You read the clean transcript and it’s… not great. It rambles. Sentences start and change direction halfway through. There are “so”s and “basically”s and “you know what I mean”s everywhere.

That’s normal. Speech and writing are different things, and a transcript is speech in written form. It needs editing, but editing is so much faster than writing, because the hard part (figuring out what to say and in what order) is already done.

When we turn a transcript into a post, we go through it in roughly three passes.

Pass one: cut

Delete everything that only makes sense with video. “As you can see here.” “Let me just move this over.” “Smash that like button.” Delete the filler words. Delete the false starts, the bits where you began a sentence, stopped, and began it again.

You’ll often cut a third of the words in this pass. That’s fine. A 1,500-word transcript becoming a 1,000-word post is a good result.

Pass two: shape

Videos are often structured for watching, not skimming. Readers skim. So add the things a reader needs:

• Headings where the topic changes. If you used chapters in the video, they’re a ready-made outline.
• Lists wherever you said “there are three things” and then listed them in a long, winding sentence.
• A tighter opening. Video intros are built to keep people from clicking away. Blog intros should get to the point faster.

Pass three: write the bits video handled visually

Anything you showed instead of said is missing from the transcript. If you held up two products and said “this one versus this one”, the reader has no idea which is which. Fill those gaps with a sentence or an image.

This pass is also where you add links: the products you mentioned, the earlier video you referenced, the source for that statistic you quoted.

Other things a transcript is good for

A blog post is the obvious use, but not the only one.

• Show notes and descriptions. Skim the transcript for the key points and you’ve got a much better video description than “In this video I talk about…”.
• Newsletters. One strong section of a video often makes a complete newsletter on its own.
• Quotes for social posts. The best lines from a video are easy to find when you can search for them instead of scrubbing through footage.
• Searching your own back catalogue. Keep plain-text transcripts of every video in a folder and you can find “that video where I mentioned the cheap microphone” in seconds.
• Translations. A clean transcript is a much easier starting point for translating a video, whether you do it yourself or work with a translator.

Need a different format instead?

Sometimes you don’t want plain text at all. You just need the captions in another format: web players and some editors want VTT, while the file you have is SRT. Going from SRT to VTT is mostly a matter of adding the header and swapping commas for periods in the timecodes, and the Subtitle Cleaner has a toggle for exactly that.

Start with one video

Pick a video that did well, one where the comments suggest people found it genuinely useful. Pull the captions, clean them, and give yourself 45 minutes to edit the result into a post.

You’ll probably find it’s the fastest blog post you’ve ever written. You wrote most of it months ago. You just said it out loud instead of typing it.