SRT Caption Guide: Subtitle Formats, Timing, and Accessibility

Captions are no longer optional — YouTube, Netflix, and most social platforms require them for accessibility. Whether you're a YouTuber, course creator, or filmmaker, understanding subtitle formats and timing is essential.

Common Subtitle Formats

SRT Format Structure

Characters Per Second (CPS)

CPS measures reading speed. If captions appear too fast, viewers can't read them:

• 15 CPS — Comfortable for most viewers
• 17 CPS — Maximum recommended
• 20+ CPS — Too fast — split into two lines or extend duration

Common Issues and Fixes

• Timing drift — Apply a time offset to shift all captions forward or back
• Long lines — Wrap at 42 characters per line, max 2 lines
• Format conversion — SRT ↔ VTT conversion is straightforward
• Auto-generated errors — YouTube's auto-captions often have timing and accuracy issues

Worked Example: Fixing a 2.5-Second Drift and a Fast Cue

A 12-minute talking-head video was cut to remove a 2.5-second intro, but the captions were never updated — every cue now appears 2.5 seconds early. Fixing it:

1. Load the SRT into a formatter and apply a +2.5s offset to all cues: 00:00:01,000 becomes 00:00:03,500
2. Scan for cues over the CPS limit. One cue holds 68 characters across 2.5 seconds — 27 CPS, far above the 17 CPS ceiling
3. Split it into two cues of 34 characters each, 2.5 seconds apiece — now 13.6 CPS, comfortable to read
4. Wrap any line over 42 characters, then renumber the cues sequentially

Total time for the whole file: about five minutes with the SRT Caption Formatter, versus an hour of nudging individual cue timings by hand in a text editor.

Common Mistakes

• Editing cue numbers but not timestamps — the sequence numbers are just labels; players order cues by timestamps. Renumbering without fixing times changes nothing on screen.
• Creating overlapping cue times — two cues occupying the same milliseconds render stacked or flicker. Leave a small gap of 50-100ms between consecutive cues.
• Using periods instead of commas in SRT — SRT wants 00:00:01,000 while WebVTT wants 00:00:01.000. Copying the wrong one breaks parsing on strict players.
• Leaving one line of 90 characters — viewers read the start, lose the end, and give up. Hard-wrap at 42 characters per line, two lines maximum.
• Uploading captions without a final watch-through — auto-generated timing errors, missing punctuation, and unmarked speaker changes are obvious when reviewing at 1.5× playback and invisible in the raw file.

Frequently Asked Questions

Should I burn captions into the video or use SRT?

Use SRT whenever the platform supports it. Side-loaded captions can be toggled, indexed by search, translated, and fixed later. Burned-in text is permanent, hurts aspect-ratio flexibility, and cannot be extracted for SEO.

How do I convert SRT to VTT?

Change the header to WEBVTT, replace the comma in each timestamp with a period, and delete the sequence numbers (they are optional in VTT). Most converters do this in one click, and the result plays in any HTML5 video element.

What CPS should I target for accessibility?

Aim for 15 CPS or less. Guidance for caption readability suggests viewers should not have to rush; 20+ CPS forces skim-reading, which defeats the purpose for deaf and hard-of-hearing viewers who rely on captions entirely.

Why do my captions look fine locally but misaligned on YouTube?

YouTube re-processes uploaded SRT files and can shift timing by a frame or two, and preview playback can differ from the published version. Always re-watch a minute of the published video at the start, middle, and end before calling it done.

The Bottom Line

1. SRT is the most universal subtitle format
2. Keep CPS under 17 for readable captions
3. Max 2 lines, 42 characters per line
4. Always check auto-generated captions for timing and accuracy errors

Disclaimer: This guide is for informational purposes only.