TXT, SRT, VTT and HTML: choose a transcript format
Use TXT when you have reliable words but no reliable timing. Use SRT or WebVTT when reviewed cues already match the recording. HTML is the readable page output, not another subtitle file to rename and upload. A file extension does not establish accuracy: open the file, check the text and compare its timing with the actual video.
One recording, three inputs
These downloads accompany our original English demonstration. They contain the same narration; the timed formats preserve the cue boundaries from that recording. The voice was generated locally with Windows speech synthesis. No transcription provider produced these files.
- Download TXT: words without cue timing.
- Download SRT: numbered cues with comma-separated milliseconds.
- Download WebVTT: a
WEBVTTheader and dot-separated milliseconds. - Read the actual published WordPress HTML: a text download, not a runnable WordPress page.
The files are deliberately small. Use them on a test page to learn the import workflow before working on a long interview or training recording. They are examples for this recording, not captions to reuse on another video.
Inspect the same first passage in TXT, SRT and VTT
TXT
Your video has something worth sharing. Give the words a place on the page, so visitors can watch or read at their own pace.SRT
1
00:00:00,582 --> 00:00:08,910
Your video has something worth sharing. Give the words a place on the page, so visitors can watch or read at their own pace.WebVTT
WEBVTT
1
00:00:00.582 --> 00:00:08.910
Your video has something worth sharing. Give the words a place on the page, so visitors can watch or read at their own pace.Compare the structures
| Format | Useful information | What you still need to decide |
|---|---|---|
| TXT | The transcript wording | Paragraphs, speakers and any chapter times |
| SRT | Wording plus start/end times | Whether the cues are accurate and readable |
| WebVTT | Wording plus timing in a browser-oriented track format | Whether your player exposes a usable captions track |
| Visible HTML | Text people can read, select and search on the page | Whether it is public, accessible and faithful to the recording |
A timing line is not part of the speech. A cue number is not a heading. Formatting markup is not permission to execute HTML. Check what the importer actually retains rather than assuming all track features become page features.
Import, adopt, then publish
In VidSEO 2.0, an import creates a proposed text candidate. Inspect it before adoption: recognizable accented characters, the first and last sentences, paragraph boundaries, and any timing information used for navigation. Save your working document and explicitly publish the reviewed version. A successful upload alone does not change the public page.
Observed in the official Free 2.0.0 test: all six English/French imports succeeded. Each TXT produced three segments with no start or end time. Each SRT and VTT preserved the three exact cue boundaries and the supplied wording. Preview did not alter the working draft; adoption saved a draft without exposing the transcript publicly. See the dated evidence record.
For a long file, compare a cue near the beginning, one in the middle and one near the end with playback. Check both the intended start and the end; an apparently correct first cue can hide later drift. If the recording has been edited since the subtitle file was produced, repair the timing against that new version.
Fix the common mismatches
If accents look wrong, resave the source as UTF-8 and inspect it before importing again. If timing is unavailable, keep honest plain text instead of assigning evenly spaced but invented timestamps. If speakers are ambiguous, add labels only after listening. If a visual instruction is never spoken, add a clearly identified descriptive passage rather than silently attributing it to the speaker.
Importing a timed transcript does not by itself prove that closed captions are available in every player mode. Check the actual captions control and the public page separately. The accessibility guide explains the needs covered by each representation.
Verify the result outside the editor
Open the published host page while signed out. Search for an exact sentence, inspect the HTTP response HTML, and test a timed jump if one is present. Compare those observations with the before/after exports. For editorial structure and contextual additions, continue with turning a transcript into a useful page.