SUBTITLES · Written and reviewed by the Media2Text team · Published November 3, 2026 · 9 minute read

How to create SRT subtitles that people can actually read

An SRT file is a plain-text list of numbered captions with start and end times. Creating one used to mean typing every line and timing it by hand. Today you can generate a timed transcript automatically and spend your effort where it matters: making the subtitles accurate and comfortable to read.

This guide explains how SRT works, how to create a file from audio or video, what makes subtitles readable and how to bring them into common video tools.

What an SRT file looks like

SRT, short for SubRip Subtitle, is one of the simplest and most widely supported subtitle formats. Each caption, called a cue, has four parts:

1. A sequence number, starting at 1.
2. A time range in the form hours:minutes:seconds,milliseconds --> hours:minutes:seconds,milliseconds. Note the comma before the milliseconds.
3. One or more lines of caption text.
4. A blank line that ends the cue.

1
00:00:01,200 --> 00:00:04,000
Welcome back to the channel.

2
00:00:04,300 --> 00:00:07,800
Today we're looking at three ways
to record clearer audio at home.

Because SRT is plain text, you can open and fix it in any text editor. Save it with UTF-8 encoding so accented letters and non-Latin scripts display correctly. Most problems with "broken" SRT files come from a wrong encoding, a missing blank line or a period used instead of a comma in the timestamp.

How to create an SRT file from audio or video

1. Use the final edit. Subtitles must match what viewers will hear. Transcribe the locked cut of your video, or the final mixed audio exported from the same timeline, so you do not have to re-time cues after editing.
2. Upload the file. In Media2Text, upload the video or audio. For supported video formats such as MP4, MOV and WEBM, the audio track is extracted in your browser and the original video is not uploaded. Audio files such as MP3, WAV and M4A work directly. See video to SRT and audio to SRT for format details.
3. Correct the words first. Fix names, brand terms and numbers in the transcript view, where find and replace and clickable timestamps make it fast. Editing words before export is much easier than fixing them cue by cue later.
4. Export SRT. Choose SRT in the export menu. Word-level timestamps are used when the transcription engine returns reliable word timing; otherwise cues follow segment timing. Decide whether speaker names belong in the captions; in most videos they do not.
5. Watch it back. Import the file into your editor or player and watch the opening, the busiest dialogue and the ending at normal speed.

What makes subtitles readable

Accurate words are only half of good subtitles. Viewers need enough time to read each cue while still watching the picture. Broadcasters and streaming services publish detailed style guides; the principles below are common to most of them.

Keep lines short

Use one or two lines per cue. Many guidelines suggest roughly 32 to 42 characters per line for Latin-script languages. Longer lines force the eye to travel across the screen and are harder to read on phones.

Give each cue enough time

Very short cues flash past before anyone can read them, and very long ones linger after the speaker has moved on. A common range is from about one second to around six or seven seconds per cue. Fast speakers produce dense cues; split them at natural pauses rather than squeezing a full sentence onto screen.

Break lines at natural points

Break between phrases, not in the middle of them. "Today we're looking at three ways / to record clearer audio" reads better than "Today we're looking at three / ways to record clearer audio". Avoid ending a line on an article or preposition when you can.

Match the speech, not the script

If the speaker departed from the script, the subtitles should follow what was actually said. Viewers who can hear the audio notice mismatches immediately.

Consider accessibility

Subtitles for viewers who are deaf or hard of hearing often include important sound information such as [music], [laughter] or [door slams], and identify off-screen speakers. Speech recognition captures spoken words only, so add these descriptions manually where they matter. The W3C captions guidance is a useful reference.

SRT or VTT: which should you choose?

If you are unsure, export SRT first; it is accepted almost everywhere. Media2Text can export both from the same transcript, so you do not need to choose permanently. To convert a file you already have, use the free SRT to VTT converter. The W3C WebVTT specification documents the VTT format in full.

Bringing subtitles into your video tools

Exact menus change between versions, so check the current documentation of your tool. In general:

• YouTube accepts uploaded subtitle files, including SRT and VTT, in the subtitles section of YouTube Studio. Uploaded files replace automatic captions for that language when you publish them.
• Desktop editors such as Adobe Premiere Pro and DaVinci Resolve can import SRT files onto a caption or subtitle track, where you can adjust styling and burn captions into the picture if needed.
• Web players built on the HTML video element load VTT files through a track element.
• Social platforms differ: some accept caption files, while others generate their own captions or expect burned-in text.

Separate or burned-in? A separate SRT or VTT file lets viewers switch captions on and off and lets platforms index the text. Burned-in captions become part of the picture, which is common for short social videos watched without sound. Media2Text produces the subtitle file; burning captions into video is done afterwards in your editor.

Speaker names, music and other special cases

Should subtitles show who is speaking?

Usually not when the speaker is on screen; viewers can see who is talking. Identify speakers when they are off screen, when several people talk in quick succession or when the video is an audio-only podcast with a static image. A short label such as "HOST:" or a dash at the start of a new speaker's line is enough. If you export from a transcript with speaker labels, decide before export whether to include them, because removing them cue by cue later is tedious.

Songs and background music

Speech recognition may transcribe sung lyrics or produce fragments when music is loud. For captions, it is common to write [music] or a short description instead of partial lyrics, unless the lyrics matter to the story. Remember that lyrics are often copyrighted.

On-screen text

Captions that sit on top of existing titles or lower thirds are hard to read. Most editors let you move a subtitle track, and VTT supports positioning on the web. Speech recognition does not read on-screen text, so translate or transcribe important titles separately if your audience needs them.

Very fast speech

When a speaker talks faster than viewers can read, you have two options: keep every word and accept short reading time, or condense slightly while keeping the meaning. Accessibility guidelines generally favour keeping the speaker's words where possible; marketing clips often condense. Choose one approach per video and apply it consistently.

Subtitles in other languages

To reach viewers in another language, transcribe the original first, then create a translated transcript track. The translation keeps the original timestamps, so you can export subtitles in the new language that line up with the same moments. Translation does not dub audio, and translated lines can be longer than the original, so re-check reading speed and line breaks in busy passages.

Common SRT problems and how to fix them

All subtitles are early or late by the same amount

The file is usually fine; the video simply starts at a different point, for example after a slate or intro that was not in the transcribed audio. Shift the whole subtitle track in your editor rather than editing every timestamp.

Subtitles drift further out of sync over time

This suggests the subtitles were made from a different edit or a file with a different frame rate. Transcribe the exact final version you are publishing.

Strange characters instead of accents

Re-save the file as UTF-8. Older tools sometimes default to a regional encoding.

The platform rejects the file

Check for a period instead of a comma in SRT timestamps, missing blank lines between cues or an empty file. Opening the SRT in a plain text editor usually reveals the problem quickly.

A name is wrong everywhere

Correct it once with find and replace in the transcript and export again, instead of fixing each cue by hand.

SRT subtitle checklist

1. Transcribe the final edit or the final mixed audio.
2. Correct names, terms and numbers in the transcript before export.
3. Export SRT for editors and uploads, VTT for web players.
4. Keep cues to one or two short lines with enough reading time.
5. Add sound descriptions manually if your audience needs them.
6. Import, then watch the start, the busiest passage and the end.

Ready to make a subtitle file? Start with video to SRT or audio to SRT. The free trial covers up to 2 lifetime transcription tasks sharing 75 minutes in total, with no credit card.