When an SRT file will not load, shows garbled text or plays at the wrong moments, the cause is almost always one of a short list of problems: the file name, the extension, the text encoding, the layout of the cues, or the timings. A few platforms add their own rules on top. Each one has a quick fix, and you rarely need to make the file again.
Work through the checks in order; the first three take seconds and solve most cases. Each fix names a free tool on capsai.co that runs in your browser, so the file is never uploaded. One note before you start: CapsAI's caption generator does not import or export SRT files. These tools are separate utilities for subtitle files you already have.
Quick diagnosis
| What you see | Likely cause | Where to look |
|---|---|---|
| Player shows no subtitles at all | File name does not match the video, or the extension is wrong | Checks 1 and 2 |
| Strange symbols instead of letters | Wrong text encoding | Check 3 |
| Upload rejected or the file reads as empty | Broken cue layout or timestamps | Checks 4 to 6 |
| Some lines missing or merged | Missing blank lines or overlapping times | Checks 5 and 7 |
| Every line early or late by the same amount | Constant offset | Check 8 |
| Fine at the start, worse later | Frame-rate mismatch | Check 9 |
| Tags or brackets visible on screen | Styling or SDH markup | Check 10 |
1. The file name does not match the video
Desktop players such as VLC find subtitles by name. The VideoLAN wiki says to give the subtitle file "the same name that your movie", and that when both names match, VLC opens the subtitles automatically. So Holiday.mp4 needs Holiday.srt next to it in the same folder. A file called Holiday (1).srt or holiday_subs.srt will not load on its own.
Fix: rename the SRT to match the video exactly, apart from the extension. Or load it by hand from the player's subtitle menu, which works whatever the name.
2. The extension is really .srt.txt
Windows and macOS can hide file extensions. If you saved the file from a text editor, it may be called captions.srt.txt even though you see captions.srt. Players and upload forms then treat it as a text file.
Fix: turn on file extensions in your file manager, rename the file so it ends in .srt only, and confirm when asked to change the extension. If you are on Facebook, the name has a further rule, covered in check 11.
3. The text encoding is wrong
If you see é, ’ or the replacement symbol (a question mark in a diamond) where accented letters, quotes or non-Latin scripts should be, the encoding is the problem. Wikipedia's SubRip article notes there is no official encoding standard for SRT files, and the original SubRip program defaulted to Windows-1252, which Windows often labels ANSI. Platforms are stricter: YouTube says the file "must be in plain UTF-8", and Facebook's caption help says the same.
Fix: open the file in a text editor and save it again as UTF-8. If the text is already damaged, save from the original source rather than the broken copy. A byte order mark (BOM) usually helps a reader detect UTF-8, but if a strict tool skips your first cue or shows a stray character before the number 1, try saving as UTF-8 without BOM.
To check the result, paste the text into the subtitle encoding checker. It flags replacement characters, hidden null bytes, a BOM, Unicode normalization and line endings. It cannot tell for certain which encoding the original bytes used, but it shows whether the text you have is clean.
4. The timestamps use a period instead of a comma
SRT writes a comma before the milliseconds: 00:00:01,200. WebVTT uses a full stop: 00:00:01.200. A VTT file renamed to .srt, or an SRT edited by hand, often ends up with full stops, and strict readers then reject the whole file. Facebook's help says each time should contain "only two colons and one comma".
Also look at the very first line. A file that starts with WEBVTT is a VTT file, not an SRT, even if the extension says otherwise. WebVTT was based on parts of SRT but is not fully compatible with it.
Fix: replace the full stops before the milliseconds with commas, and delete any WEBVTT header and notes above the first cue.
5. Hours or digits are missing
Every SRT time needs all its digits, hours included: 00:01:05,050, not 1:05,050 or 00:01:05,5. Facebook's help says times must "include every digit in the format hh:mm:ss,fff". WebVTT allows the hours to be left out, so this problem also comes from converted VTT files.
Fix: pad every time to two digits for hours, minutes and seconds and three for milliseconds. If the only problems are full stops or missing hours, open the file in the SRT editor and download it again: it reads either style and writes every time out in full, with a comma before the milliseconds.
6. Cue numbers are missing or out of order
Each cue starts with a number. Facebook asks for a unique sequence number for each subtitle, starting at 1. Gaps, repeats or a missing number are common after you delete or paste cues by hand, and some readers stop at the first bad cue.
Fix: renumber from 1. The SRT editor is the easiest way: open the file, edit any cue's text or start and end times, and download it. The cues come out numbered in order.
7. Blank lines are missing, or there are too many
A cue is a number, a timing line, one or two lines of text, then one blank line. If the blank line is missing, two cues run together. If a text line is left empty in the middle of a cue, the reader thinks the cue has ended. Facebook's help asks for blocks separated by a single blank line, and no spaces or lines after the last block.
Fix: make sure each cue ends with exactly one empty line, remove empty lines inside cues, and delete trailing blank lines at the end of the file.
8. Times overlap, run backwards, or the whole file is shifted
Three related timing faults:
- Overlaps: a cue starts before the previous one ends. Some players hide one of them.
- End before start: a typing slip such as 00:00:05,000 --> 00:00:04,000. The cue may never show.
- A constant offset: every line is early or late by the same amount, usually because an intro was added or trimmed.
Paste the file into the subtitle sync checker. It lists overlapping cues, reversed timestamps and captions held on screen for more than ten seconds, by cue number, so you can fix each one in the editor.
For a constant offset, use the subtitle time shift tool. Enter a positive or negative number of milliseconds, and it moves every cue by that amount. To find the number, note when the first word is spoken and compare it with the first timestamp. If a shift would push a time below zero, the tool sets it to 00:00:00,000 instead of writing a negative time.
9. The subtitles drift further out of sync over time
If the first minute is fine but the end is badly off, a single shift will not fix it. The usual cause is a frame-rate mismatch: the subtitles were timed for one version of the video, such as 23.976 frames per second, and you are playing another, such as 25. That version runs about 4 percent faster, so by the one-hour mark the subtitles are about two and a half minutes late.
Fix: use the subtitle frame rate converter. Pick the frame rate the subtitles were made for and the frame rate of your video, and it scales every time by the difference. If the drift does not match a common frame-rate pair, the cause may be an edit in the middle of the video instead. In that case, shift each section separately.
10. Tags, brackets or speaker names show up on screen
SRT has no official styling. Some players read simple tags such as italic, but YouTube says it recognizes no style markup in SRT files, so tags can be ignored or shown as text depending on the player. Files made for deaf and hard-of-hearing viewers (SDH) also carry sound descriptions such as [door slams] and speaker labels such as JOHN:, which you may not want.
Fix: run the file through the SRT cleaner. It removes HTML-style tags, curly-brace position tags, bracketed sound descriptions, upper-case speaker labels and music notes, drops any cues left empty, and renumbers the rest. Each option can be switched off if you want to keep something.
11. The platform rejects the upload
Platforms add their own rules. As of 21 October 2026, their help pages say:
- YouTube: SRT is a supported basic format, but only basic files are accepted, no styling is recognized, and the file must be plain UTF-8. In YouTube Studio, choose Upload file, then With timing for an SRT, because the file already has its timings.
- Facebook: captions are added from the post's Edit menu with a SubRip (.srt) file. Facebook's naming help gives the pattern filename.[language code]_[country code].srt, for example filename.en_US.srt, with the language code in lower case and the country code in upper case. Its reel-caption help lists the rules in checks 3 to 7: numbers from 1, every digit in hh:mm:ss,fff, UTF-8, single blank lines and nothing after the last block.
- LinkedIn: LinkedIn's help says you attach an SRT file with Select Caption before the video is posted. The help page describes this in the desktop experience.
When an SRT file is not the right tool
An SRT is a separate track that viewers can switch off. That suits YouTube, course sites and video editors, but not apps where you cannot attach a caption file, or captions that must look the same everywhere. For those, the captions need to be part of the picture. CapsAI does this from the video itself, not from an SRT. It makes captions from the speech in 11 spoken languages, lets you edit the words and timings and choose a caption style, and exports an MP4 with the captions burned in. It does not read your SRT file, and it does not give you one back.
What to do next
Start with the three fast checks: the file name matches the video, the extension is only .srt, and the file is saved as UTF-8. If the file still fails, check the timestamps and the cue layout, then run the sync checker before you upload again. Fix single cues in the SRT editor.
If you would rather skip caption files for short videos, you can upload an MP4 or MOV up to 10 minutes long to CapsAI and download it with the captions burned in. New accounts get 3 free minutes to start, and one-time packs start from $4 (₹49 in India, €4 in the euro area).
