How to convert audio to SRT subtitles
You've got a video or a recording, and you need subtitles. Maybe you're posting on YouTube, maybe it's a course, maybe you just want captions on a clip because half your audience watches on mute. Whatever the reason, the file you usually need is an SRT.
It sounds technical, but an SRT file is just text with timestamps. Once you understand that, making one from audio is straightforward. This guide covers how it works, how to make one, and the small fixes that separate good captions from annoying ones.
What is an SRT file?
SRT stands for SubRip Subtitle. It's a plain text file, and every subtitle in it has the same simple structure:
1
00:00:01,000 --> 00:00:03,500
Welcome back to the channel.
2
00:00:03,800 --> 00:00:07,200
Today we're going to talk about
how to record better audio.
Each block has four parts:
- A number. Just a counter: 1, 2, 3 and so on.
- A time range. When the subtitle appears and when it disappears, written as hours:minutes:seconds,milliseconds. Notice the comma before the milliseconds. That's a quirk of the SRT format.
- The text. One or two lines of what's being said.
- A blank line. This separates one subtitle from the next. Leave it out and some players get confused.
That's the entire format. You can open an SRT in any text editor, including Notepad, and read or change it by hand.
SRT or VTT?
You'll also see a format called VTT (WebVTT). It's very similar, but it starts with the line WEBVTT, and it uses a full stop instead of a comma in the timestamps (00:00:01.000). Websites that play video with the HTML player tend to prefer VTT, while YouTube, many video editors and most social tools happily take SRT.
If you're not sure, SRT is the safer default. Almost everything reads it. If a platform rejects it and asks for VTT, you can usually get the other format from the same transcription.
Why you can't just use a normal transcript
A plain transcript is only the words. It doesn't know when each word was spoken. Subtitles need timing, so the software has to work out where each sentence starts and ends in the audio. That's called alignment, and speech recognition models do it as a by-product of transcribing.
That's why you can't convert a text document into an SRT by itself, but you can turn audio into one directly. The tool listens, writes the words, and notes the timings at the same time.
Step by step: audio to SRT
1. Start with the best audio you have
If your subtitles are for a video, use the video file or the exported audio from your editor. Clean audio gives you cleaner text and, just as importantly, cleaner timing. Music underneath the voice, echo and background noise all make alignment less precise.
2. Transcribe the file
Open the audio to text tool, add your file and start the transcription. MP3, M4A, WAV and video files all work, and the language is detected on its own. If you're working from a phone recording, voice memo to text is a good starting point.
3. Check the text before you export
This is the cheap moment to fix mistakes. Correcting a misspelled name in the transcript is much easier than finding it later inside the subtitle file. Play the audio and read along, fixing names, numbers and jargon.
4. Export as SRT
Use the SRT option in the transcript toolbar. The timestamps go into the SRT file; the TXT download stays as plain text without timings. Keep the file name the same as your video, for example my-video.srt next to my-video.mp4. Many players load subtitles automatically when the names match.
5. Test it
Don't skip this. Load the SRT with the video in a player such as VLC, or upload it to your platform, and watch a couple of minutes from the start and from the middle. You're checking that:
- Subtitles appear at the right moment, not early or late.
- Lines are not too long to read.
- Nothing runs off the screen.
Making subtitles easy to read
A technically correct SRT can still be painful to watch. These rules of thumb come from how professional captioners work, and they make a real difference.
Keep lines short. About 40 to 42 characters per line is a common limit, and two lines per subtitle is the maximum most people can read comfortably. If the subtitle is taller than that, split it.
Give people enough time. A subtitle should usually stay on screen for at least a second, and no longer than about six or seven seconds. A good target is reading speed of roughly 15 to 20 characters per second. If it flashes by, nobody can read it.
Break at natural points. Split lines where you'd pause when speaking, not in the middle of a phrase. "the red / car" is awkward. "the red car / was parked outside" reads smoothly.
Don't leave gaps that feel like glitches. If a speaker pauses briefly, the subtitle can just stay up. Flickering on and off every half second is distracting.
Keep it faithful. For captions, write what was said. Remove heavy filler words if they clutter the screen, but don't rewrite the meaning.
Automatic tools can split lines sensibly, but they don't always follow these rules perfectly. Spend a few minutes on a spot check rather than assuming it's fine.
Fixing common SRT problems
The subtitles drift out of sync
If they start on time and slowly get later or earlier, the video may have a different frame rate or have been edited after you made the audio. Re-export the audio from the final video and transcribe that. If the whole file is a fixed amount off (say, two seconds late), many players let you shift subtitle timing, and subtitle editors can shift everything at once.
The text appears with strange symbols
This is usually an encoding issue. SRT files should be saved as UTF-8. If a player shows odd characters, particularly with accented letters or non-English text, open the file in a text editor and re-save it as UTF-8.
The platform rejects the file
Check that the structure is intact: numbers in order, a blank line between each block, and the arrow --> between times. One missing blank line can break an upload. Some platforms also want VTT instead.
A name or word is wrong everywhere
Open the file in a text editor and use find-and-replace. It's plain text, so this takes seconds.
Uploading to common platforms
- YouTube. In YouTube Studio, open the video, go to Subtitles, add the language, choose upload file and pick the SRT. Check the timing and publish.
- Social platforms. Many let you upload an SRT file when you post a video. Some prefer to burn captions in, so check what your platform supports.
- Video editors. Most editors can import an SRT as a subtitle or caption track, and you can then style the text.
- Course platforms and LMSs. Most accept SRT or VTT in the video settings.
If you want captions permanently embedded in the picture, you'll need to "burn them in" using a video editor. An SRT file on its own stays separate, which is nice because viewers can turn it on or off.
Subtitles in more than one language
Once you have a clean SRT in one language, you can translate the text and keep the timings. The important thing is to keep the structure intact: numbers, times and blank lines stay exactly as they are, and only the text lines change. Translated lines are often longer than the original, so recheck line length afterwards.
Do you really need SRT for SEO?
Subtitles and transcripts help people, and they help video platforms understand what's in the content. But the biggest reason to bother is simpler: a lot of people watch with the sound off, and people who are deaf or hard of hearing need captions. Treat any search benefit as a bonus.
The short version
An SRT is a text file of numbered, timed subtitles. To make one from audio, transcribe the file, fix the text, export as SRT, and test it in a player. Keep lines short, give people time to read, and check the start and middle of the video before you publish.
Ready to make one? Open the audio to text tool, drop in your file, and export your subtitles once the transcript looks right.