A subtitle file that works in VLC and does nothing in a browser is not a mystery. Both programs are reading the same bytes and reaching different conclusions, because one of them is stricter than the other about something specific. Finding out which thing is the whole job.
The frustrating part is that subtitle failures are usually silent. A malformed MP4 throws an error. A malformed subtitle file usually produces a video that plays perfectly with no text, and no message anywhere telling you why. This guide covers the defects that cause that, in rough order of how often they occur.
1. The file is the wrong format for the target
This is the most common cause and the easiest to miss, because nothing about the file is broken. A browser loading subtitles through an HTML5 track element accepts WebVTT only. Hand it a perfectly valid SRT file and it fails silently — no console error, no warning, just no subtitles.
The same is true in the other direction: a large number of broadcast and authoring tools accept SRT only. The file is fine; the target does not read that format.
2. A byte order mark at the start of the file
A BOM is an invisible character at the very beginning of a text file. It is three bytes (EF BB BF) that editors on Windows sometimes add to signal "this file is UTF-8". It is legal in some formats and a defect in others, and subtitle parsers vary wildly in whether they tolerate it.
What happens when they do not: the parser reads the first line as "\uFEFF1" instead of "1", decides the file does not start with a cue number, and either skips the first cue or rejects the whole file. In a player that keeps it, the BOM shows up as a stray character before the first line of text.
A BOM is invisible in most editors, which is why people spend an hour looking for a problem they cannot see. The only reliable way to find it is to check the first bytes of the file, or to open it in an editor that displays the encoding explicitly.
3. Mixed line endings inside one file
Text files end their lines with one of three things: a line feed (LF, what Unix writes), a carriage return plus line feed (CRLF, what Windows writes), or occasionally a bare carriage return (CR, what classic Mac OS wrote). Every parser accepts LF and CRLF.
The problem is a file that uses both. This happens when a file is edited by two different tools, or when text is pasted from one application into another. The result is not a clean failure — it is a file where some cues work and others appear to have lost their text, because the parser split the block in the wrong place.
This one is worth checking early, because it produces the confusing symptom of a file that is "partly broken".
4. Missing or non-sequential cue numbers
SRT requires a number above every cue. Some parsers are strict about it and some are not, which is exactly what makes this defect so inconsistent: a file missing its numbers works in one player and shows nothing in another.
A subtler version is numbers that are present but not sequential — 1, 2, 3, 5, 6 after a cue was deleted. Most players do not care. Some do, and the failure looks like the cues after the gap are missing.
Converting from WebVTT makes this more likely than people expect, because WebVTT cue identifiers are optional and may be arbitrary text. A conversion that copies the identifier "intro" into the number slot produces a file whose first block starts with a word, not a digit.
5. Timestamps that do not parse
Subtitle timestamps are unforgiving. The defects that occur in practice:
- A period where SRT wants a comma, or the reverse. This is the single most common timestamp defect.
- A missing or doubled space around the arrow.
- Two-digit hours instead of two, or a missing hour field where the format requires one.
- Milliseconds written with two digits or four instead of exactly three.
- Minutes or seconds above 59, which some strict parsers reject even though the arithmetic is unambiguous.
A file with one unparseable timestamp often fails as a whole rather than dropping one cue, because the parser loses its place in the structure and cannot recover.
6. A cue that ends before it starts
An out-time earlier than the in-time is invalid in every format and produces different behaviour everywhere: some players show the cue for zero frames, some show it forever, some reject the file. It usually comes from a manual edit or from a tool that did arithmetic on the timestamps without validating the result.
7. Overlapping cues
Two cues that cover the same moment is not strictly illegal — some formats and players handle it by stacking both lines on screen — but it is almost never intended, and in formats that allow only one cue at a time the second one is dropped or the file is rejected.
Overlaps usually appear after a timing shift, a merge, or a split operation. They are a sign that the timing was modified without checking the result.
8. Empty cues
A cue with a timestamp and no text is a block with nothing in it. Some players skip it, some reserve the space and show nothing, some treat the following cue as part of the empty block. It is a frequent artefact of automated processing, and it is easy to spot once you know to look for a blank line where text should be.
9. The wrong character encoding
This one does not make the file fail to load — it makes it load with gibberish. A file saved as Windows-1252 and read as UTF-8 shows accented characters as pairs of odd symbols; a file saved as UTF-8 and read as Windows-1252 shows the same in reverse.
It is worth separating from the other eight because the symptom is different: the subtitles appear, at the right times, in text nobody can read. That is an encoding problem, not a structural one, and re-saving the file as UTF-8 fixes it.
Working out which one you have
The symptoms point at the cause more reliably than people expect:
| Symptom | Most likely cause |
|---|---|
| No subtitles at all, no error anywhere | Wrong format for the target, or a BOM |
| First cue missing, the rest fine | BOM |
| Some cues work, others show no text | Mixed line endings |
| Everything after a certain point is missing | A non-sequential cue number, or an unparseable timestamp |
| Text appears but is unreadable symbols | Wrong character encoding |
| Two lines on screen at once | Overlapping cues |
| The same file works elsewhere | A parser that is stricter than the spec, not a broken file |
The last row is the important one. "It works in VLC" does not mean the file is correct — it means VLC is more forgiving. A file should be checked against the format, not against the most lenient program you happen to have installed.
A check that tells you what is wrong
Opening the file in an editor and reading it will find the obvious problems — a missing number, a comma where a period belongs — but it will not find a BOM, will not tell you whether line endings are mixed, and will not catch an overlap three thousand cues in.
What you want is a check that reports the problem by line number and says what is wrong, rather than a score. A score tells you the file has issues; a line number tells you where to look.
Try it on your own file — no upload, nothing leaves your browser.
Open the Subtitle Format Validator - Fix SRT/VTT Structure Errors →If the validator reports the file as clean and the player still ignores it, the problem is almost always the first item on the list — the format itself, not the contents.