CaptionPass is two things sharing an engine. It is a caption converter and repair tool for SRT and WebVTT — timing, overlaps, reading speed, encoding, the platform rules that silently reject an upload. And it is a caption conformance layer: it grades a caption file against named accessibility standards and issues a signed, independently verifiable record of what was checked, against which version of which standard, and what the result was per criterion.
It is not transcription, not speech-to-text, and not a burned-in subtitle renderer. It sits downstream of caption vendors rather than against them: they answer are there captions; this answers can we show what they were checked against.
Every figure below is computed from the registries the product actually runs on, at the moment you loaded this page. None of them is typed in — a stale hand-written traction number is worse than no number, because you have no way to know it went stale and every reason to assume somebody checked.
Where there is no honest figure, the row stays and says why.
The argument on this page is that our evidence is produced by a machine rather than by marketing. These are live, need no account, and are the fastest way to find out whether that is true.
Generated through the production engine on every page load, signed with the key the deployment actually holds. Not a fixture.
Including its failures. A sample with nothing wrong in it would show you nothing.
Including one we hold no copy of, and without trusting whoever sent it — us included.
When YouTube, TikTok and the rest changed what they accept, with the date each was detected.
The engine reads the caption file and never the audio. So whether captions match what was actually said is outside what it can determine, and it is reported as not evaluated on every receipt issued — permanently, by design. Any tool certifying caption accuracy from a sidecar file is claiming something it cannot know.
A receipt is evidence of what was checked. It is not a legal opinion, an audit, or a safe harbour.
Answer pages for the compliance questions — Four guides on evidencing ADA Title II captions, filling in the caption rows of a VPAT, the FCC's four caption quality standards, and checking captions against WCAG 2.1 AA. Every coverage claim on them is generated from the engine's own registry, so a page cannot claim the product settles something it does not.
Commercial questions — revenue, customers, pipeline — are not answered on this page, and that is deliberate rather than coy: it is public and permanent, and those figures belong in a conversation where they can be asked about properly. Enterprise enquiries reach a person.