CaptionSync
Guide 03

SRT vs WebVTT vs SBV: format differences and clean conversion

The four files people call subtitles are not interchangeable. SRT is the interchange standard, WebVTT is the web standard with settings and styling, SBV is the older YouTube format, and TXT is a transcript with no timing at all. Here is what each one stores, what survives conversion, and the encoding details that cause most platform rejections.

SRT: the safest common denominator

SubRip files are plain text blocks separated by a blank line. Each block has a sequence number, a timecode line and one or more text lines:

1
00:00:04,000 --> 00:00:06,200
I never said that.

2
00:00:06,400 --> 00:00:08,900
You did, on the phone.

The timecode uses a comma before the milliseconds, always three digits, and the arrow is two hyphens and a greater-than sign with spaces around it. Most tools tolerate extra spaces, missing sequence numbers or dot milliseconds, but a strict parser may not, so export clean files. SRT has no header, no styling and no positioning: every other feature people associate with subtitles lives in the player or in a different format.

WebVTT: the web format

WebVTT starts with a required WEBVTT signature line and a blank line, and uses a dot before the milliseconds. Cue identifiers are optional, and a cue can carry settings after the timecode:

WEBVTT

intro-1
00:00:04.000 --> 00:00:06.200 align:start position:10%
I never said that.

00:00:06.400 --> 00:00:08.900
You did, on the phone.

The settings are the reason to choose VTT. align controls horizontal alignment, position places the cue across the width, line sets the vertical position, size sets the width, and region attaches the cue to a named region. VTT also supports inline markup: bold, italic, underline, class spans, voice annotations such as <v Ana>, and ruby text for East Asian annotations. It supports comments (NOTE) and style blocks, which SRT cannot express.

VTT timestamps may omit the hours when the file is shorter than an hour, so 00:04.000 is valid. That flexibility is also a trap: a converter that assumes hours will read it as four minutes and one that assumes minutes will read it as four seconds. CaptionSync accepts both and normalises on export.

SBV: the old YouTube format

SBV puts both timecodes on one line, separated by a comma, with no sequence numbers and no header:

0:00:04.000,0:00:06.200
I never said that.

0:00:06.400,0:00:08.900
You did, on the phone.

The hour is a single digit, the milliseconds use a dot, and the separator between the two times is a comma, which is the detail that breaks naive parsers: a comma is also the SRT millisecond separator. YouTube still accepts SRT and VTT for manual uploads, so SBV is mostly legacy, but it appears in older caption exports and in tools that predate VTT support.

TXT: a transcript, not a subtitle

A plain text file has no timing, no cue boundaries and no formatting. It is what a client often sends when they have a script, a transcript from a meeting, or a rough translation. CaptionSync reads it, strips any timecode-like lines, splits paragraphs into readable chunks and assigns each chunk a duration at a calm 17 characters per second. The result is a starting point for timing, not a deliverable. The TXT export does the reverse: it joins each cue into a paragraph and drops the timing, which is useful for handing a verified script back to a writer.

What survives a conversion

What each format can store, and what is lost when you convert to it
FeatureSRTWebVTTSBVTXT
TimingYesYesYesNo
Sequence numbersYesOptionalNoNo
Cue settingsNoYesNoNo
Styling and voice tagsBasic HTML toleratedYesNoNo
Millisecond separatorCommaDotDotNone
HeaderNoneWEBVTTNoneNone
CommentsNoNOTE blocksNoNo

Conversion to a richer format cannot invent information. SRT to VTT keeps the text and timing but adds no positioning; VTT to SRT drops settings, classes and voice annotations, keeping the text. If a client asks for SRT from a styled VTT, say what will be lost rather than delivering a silently flattened file.

Encoding, line endings and the BOM

Most rejections that are not about timing are about encoding. Export UTF-8 without a byte order mark. A BOM at the start of an SRT file appears as a stray character before the first cue number, and some players show it or refuse the file. A WebVTT file with a BOM before WEBVTT fails the signature check in strict players. Old Windows tools may write Windows-1252, which turns curly quotes and accented names into mojibake when read as UTF-8.

Line endings are usually tolerated: LF is standard on the web, CRLF is standard on Windows, and most players accept either. Blank lines between cues, however, are structural. A file with Windows line endings and no blank line between blocks may parse as one enormous cue on some players. When in doubt, export from a tool that writes clean blocks and verify by opening the file in a text editor.

A conversion checklist

  1. Decide the delivery format from the platform, not from the file you received.
  2. Finish timing and readability in the working format before converting.
  3. Keep the text intact: do not let a converter strip italics unless the target cannot hold them.
  4. Check the first line for a BOM and the header for WEBVTT when delivering VTT.
  5. Check the millisecond separator: comma for SRT, dot for VTT and SBV.
  6. Open the exported file in the target player once, not only in your editor.
  7. Keep the working file; conversion is one way for settings and styling.

Convert without losing the timing

Open SRT, VTT, SBV or TXT, fix the sync and the readability, then export any of the four in one click.

Open CaptionSync

Questions about subtitle formats

Does SRT support styling?

Not officially. Some players understand simple HTML such as and inside an SRT file, and many tools write them, but positioning, fonts and colours are not part of the format. If you need reliable styling or placement, deliver WebVTT with cue settings and CSS, or burn the subtitles into the picture.

Why does my WebVTT file fail to load?

The three usual causes are a missing WEBVTT signature on the first line, a byte order mark before it, or a timecode with a comma instead of a dot. A VTT file also needs a blank line after the header and between cues. Renaming an SRT file to .vtt does not make it valid WebVTT.

Does YouTube still need SBV?

No. YouTube accepts SRT and WebVTT for subtitle uploads, and SBV is mostly a legacy format from older caption tools. If a client asks for SBV specifically, produce it, but for YouTube delivery SRT is the simpler and better supported choice.