A managed captioning vendor's QC pass is a human reviewer bundled into the price of a captioning job: they check whether the transcription is accurate, whether speaker labels and sound descriptions are used correctly, and whether a translated subtitle reads naturally. That review is valuable and this tool does not attempt to replace it — but it costs money per project, is not instant, and needs your media or caption file uploaded to that vendor's platform.
This tool checks a narrower, purely mechanical slice of caption quality — reading speed, line length, lines per cue, on-screen duration, overlapping cues, and timing — against commonly-used caption style-guide thresholds. It runs instantly, for free, from pasted text in your browser or a local file on the CLI, and nothing you check is ever uploaded anywhere.
| Question | Where to answer it |
|---|---|
| Is this caption file's transcription actually accurate? | A human review, such as a captioning vendor's QC pass — this tool has no model of meaning and cannot judge accuracy. |
| Does every cue meet a stated reading-speed cap, line-length limit, and duration range? | This checker — a deterministic, repeatable answer in seconds, free. |
| Are speaker labels, sound descriptions, or a translation's wording correct? | A human reviewer — outside what mechanical checks can verify. |
| I want this run automatically on every caption file before a human ever reviews it. | This checker's CLI — wireable into your own CI, exits non-zero on a danger-severity finding. |
| I'd rather not upload my caption file to a third party for a quick pacing check. | This checker — runs entirely in your browser or on your own machine; nothing is uploaded. |
Neither replaces the other. A vendor's QC pass is what catches whether the words are right. This tool is what catches whether the timing and formatting are right — cheaply, instantly, and as many times as you want, before or instead of paying for a human pass on the mechanical half of the job.
Different jobs, and they are often used together. A paid captioning vendor's QC pass is a human review bundled into the cost of a managed captioning job — checking whether the transcription is accurate, whether speaker labels and sound descriptions are used correctly, and whether a translation reads naturally. This tool checks a narrower, purely mechanical set of properties — reading speed, line length, on-screen duration, overlaps, and timing — instantly and for free, on a file you already have. It does not replace a human accuracy review; it catches the pacing and formatting problems a human reviewer would otherwise have to check by eye.
Anything that requires understanding meaning rather than measuring timestamps and character counts: whether the transcription is actually correct, whether a speaker change is labelled, whether a sound description is accurate, or whether a translated subtitle reads naturally in the target language. This tool has no model of meaning at all — it only measures mechanically verifiable properties of the cues you give it.
A specific, numeric, repeatable answer to whether every cue meets a stated reading-speed cap, line-length limit, and duration range — the kind of check that is tedious and error-prone for a human to do cue-by-cue by eye, and that a vendor's QC process may or may not check against the same numbers you care about. Because it is deterministic arithmetic, running it twice on the same file always gives the same answer, and you can wire it into your own CI so it runs on every caption file before a human ever sees it.
No. The browser page parses whatever text you paste in your own browser and never sends it anywhere; the CLI reads a local file directly from your own disk. A managed captioning vendor's QC pass, by contrast, necessarily involves sending your media or caption file to that vendor's platform for a human to review — that upload is a normal and unavoidable part of how a human review works, and it is one difference this free, local tool is specifically built to avoid for the mechanical checks it does run.
More: why caption quality is a different question from caption presence.