Skip to content

Caption evidence for colleges, courts and public bodies

Know which captions pass.
Prove it to anyone.

CaptionPass checks every caption file in your library against WCAG 2.1 AA, FCC 79.1, DCMP and EN 301 549, points to the exact caption behind every finding, and hands you a signed record.

Free: 2 caption hours a month of signed records, no card.

biology-101-lecture-07.srt284 captions · 28:00Sample lecture5passes6fails5needs a person
Captions

Use the up and down arrow keys to move between captions.

Caption 3 · 00:13 to 00:17

Today we move out to the edge of the cell.

DCMP presentation-ratePresentation rate

Cue 3 is shown at 170 words per minute (max 160).

1 stretch of ten seconds or more with no captions.

Drop an .srt or .vtt file on the player to check it here. It never leaves your browser.

Caption file rejected by YouTube, Canvas or Vimeo?

The free converter repairs timing and format and converts between SRT, VTT and more. 5 files a day with no sign-in, and the words are never changed.

Open the free converter

Who needs captions to be right.
And by when.

For a deaf student a wrong caption is a lecture they did not get. For the institution it is a federal standard with a date on it, and a question someone will ask: what did you check, and what did you find?

US deadlines for accessible web content and video, measured from today
  1. ADA Title II

    Takes effect 26 April 2027 — 5 months away

    State and local government serving 50,000 or more, including public universities

  2. HHS Section 504

    Takes effect 11 May 2027 — 6 months away

    HHS-funded organisations with 15 or more employees

  3. ADA Title II

    Takes effect 26 April 2028 — 17 months away

    Smaller public entities and special districts

  4. HHS Section 504

    Takes effect 10 May 2028 — 18 months away

    HHS-funded organisations with fewer than 15 employees

Accessibility and ADA coordinators

Public universities, community colleges, school districts, cities and counties in the US

The Justice Department's web rule makes WCAG 2.1 AA the standard for their web content, and the video on it, captions included. When a complaint arrives, they are the person asked to show what was checked.

WCAG 2.1 Level AA

Takes effect 26 April 2027 — 5 months away

Deaf and hard-of-hearing students, staff and residents

Every recorded lecture, board meeting and public hearing

A caption that is wrong, late or missing is a lecture they did not get. In 2015 the National Association of the Deaf sued Harvard and MIT over missing and inaccurate captions. Both settled, Harvard in 2019 and MIT in 2020, and agreed to caption their online video.

Every day the video is up

General counsel and risk offices

Any institution that publishes video

A complaint, a lawsuit or a federal review asks what a video was checked against and what was found. A captioning vendor's invoice does not answer that. Neither does a policy.

Whenever someone asks

Health systems and academic medicine

Hospitals, clinics, and universities that receive HHS funding

The Department of Health and Human Services applies the same WCAG 2.1 AA standard to the web content and apps of the organisations it funds, the video on them included.

WCAG 2.1 Level AA

Takes effect 11 May 2027 — 6 months away

Everyone who needs it, who enforces the rules, and every source

Three answers.
Never a guess dressed as a pass.

Most checkers turn every criterion green or red. Some criteria cannot be judged from a caption file, and accuracy is one of them. CaptionPass says so, and tells you exactly which ones still need a person.

  • Passes5 of 16

    The engine checked everything a caption file can show for the criterion and found nothing wrong.

  • Fails6 of 16

    The engine found a problem, and points to the exact caption and timecode.

  • Needs a person5 of 16

    A file cannot settle it. A person has to listen, and the record says so.

Every criterion, as checked on the sample lecture above. Generated by the engine when this page was built.
AnswerStandardCriterionWhat the engine found
FailsWCAG 2.1 AA1.2.2: Captions (Prerecorded)1 finding. 151.2s with no captions between cue 118 and cue 119.
FailsFCC 79.179.1(j)(2)(ii): Synchronicity5 findings. Cue 53 overlaps with cue 54.
FailsFCC 79.179.1(j)(2)(iii): Completeness1 finding. 151.2s with no captions between cue 118 and cue 119.
FailsDCMPpresentation-rate: Presentation rate18 findings. Cue 3 is shown at 170 words per minute (max 160).
FailsEN 301 5499.1.2.2: Captions (Prerecorded)1 finding. 151.2s with no captions between cue 118 and cue 119.
FailsEN 301 5497.1.2: Captioning synchronization5 findings. Cue 53 overlaps with cue 54.
Needs a personWCAG 2.1 AA1.2.1: Audio-only and Video-only (Prerecorded)Cannot be judged from the file alone.
Needs a personWCAG 2.1 AA1.2.3: Audio Description or Media Alternative (Prerecorded)Cannot be judged from the file alone.
Needs a personWCAG 2.1 AA1.2.4: Captions (Live)Cannot be judged from the file alone.
Needs a personWCAG 2.1 AA1.2.5: Audio Description (Prerecorded)Cannot be judged from the file alone.
Needs a personFCC 79.179.1(j)(2)(i): AccuracyCannot be judged from the file alone.
PassesFCC 79.179.1(j)(2)(iv): PlacementNothing found in the parts a file can show.
PassesDCMPline-treatment: Line count and divisionNothing found in the parts a file can show.
PassesDCMPspeaker-identification: Speaker identificationNothing found in the parts a file can show.
PassesDCMPsound-effects: Non-speech informationNothing found in the parts a file can show.
PassesEN 301 5497.1.3: Preservation of captioningNothing found.

How it works

From an export folder to a record anyone can check.

  1. 01

    Export the captions you already have

    Every major video platform exports caption files. Kaltura, Panopto, YuJa, Canvas Studio, Mediasite, Vimeo and YouTube all do. You send the files, not the video.

    • .srtSubRip
    • .vttWebVTT
    • .ttmlTTML
    • .dfxpDFXP
    • .sbvYouTube SBV
    • .assASS / SSA
  2. 02

    The engine reads every caption, against four standards

    Timing, reading speed, gaps, overlaps, line length and format, checked on every caption of every file. The words themselves are never changed. Here: five sample files, 1 h 44 m of captions.

    Five sample files, checked against four standards
    FileWCAG 2.1 AAFCC 79.1DCMPEN 301 549All criteria
    biology-101-lecture-07.srt28 m
    campus-orientation-2026.srt12 m
    board-meeting-october.srt24 m
    history-210-lecture-03.srt22 m
    public-hearing-budget.srt18 m
  3. 03

    A person samples the words against the audio

    Accuracy is the one thing a file cannot prove. On a paid audit a person listens to a random sample and counts every error. The library report states the sample size and the lowest accuracy it supports, never a rounded-up promise.

    Words checked by hand
    820
    Errors found
    3
    Accuracy at least, with 95% confidence
    99.0%

    A worked example. The bound is computed by the same code that writes it onto real records.

  4. 04

    A signed record, for each file and for the library

    Each record names the file by its fingerprint, the standards and their versions, the date, and every finding. Anyone you give it to can check it on our verify page. If a single character changes, it fails.

    {
      "receiptId": "rec_sample_biology101lectur",
      "issuedAt": "2026-11-01T09:00:00.000Z",
      "file": {
        "name": "biology-101-lecture-07.srt",
        "sha256": "1f5359390c3afbdf22c667e7aa5e3e7fb0cd3d662500ad98604389d6ce17f229",
        "bytes": 21597,
        "cueCount": 284,
        "durationMs": 1671001,
        "mediaDurationMs": 1680000
      },
      "engine": {
        "engineVersion": "0.2.5",
        "presetSchemaVersion": "2",
        "runtime": {
          "node": "22.23.5",
          "icu": "78.3",
          "unicode": "17.0"
        }
      },
      "standards": [
        {
          "standardId": "wcag-2.1-aa",
          "evaluatedClausesPassed": false,
          "notEvaluatedCount": 4
        },
        {
          "standardId": "fcc-79.1",
          "evaluatedClausesPassed": false,
          "notEvaluatedCount": 1
        },
        {
          "standardId": "dcmp",
          "evaluatedClausesPassed": false,
          "notEvaluatedCount": 0
        },
        {
          "standardId": "en-301-549",
          "evaluatedClausesPassed": false,
          "notEvaluatedCount": 0
        }
      ],
      "signature": {
        "algorithm": "HMAC-SHA256",
        "keyId": "rk1",
        "value": "c9e69438cdf14c8298d78081…"
      }
    }

Published prices, in caption hours.

Start with a free account: 2 caption hours a month of your own files, with a signed record for each. If the record is useful, audit the whole library, then keep it current with Program. Captioners and production teams have Solo and Team.

For institutions

Colleges, universities, public agencies and companies. Fixed prices, one invoice, and a person sampling accuracy.

The audit

$2,500

Fixed price, up to 250 h (about 500 files of 30 minutes)

Your whole library, checked by the engine, with accuracy sampled by a person.

  • A signed record for every file
  • A library report you can hand to counsel or procurement, with the caption rows of your ACR
  • Measured accuracy on a sample, with its sample size stated
  • One invoice

$8 an hour above 250 h, up to 1,000 h; past that, a quote.

Program

From $3,000 a year

Your team checks new files as they arrive, and the report stays current.

Every size includes

  • The first audit of your library
  • Accuracy sampled by a person, one batch a year
  • The library report and your ACR's caption rows, kept current
  • Invoice or purchase order
  • Up to 10 seats, with roles
Sized by the hours of captions you check a year
Hours a yearPrice a year
150 h$3,000
500 h$6,000
1,500 h$12,000

In the first year, the included audit covers the library you start with, up to 250 h. Program's yearly hours cover the files you add after it.

Contract. For more than 1,500 h a year: systems, multi-campus institutions, broadcasters and platforms. Priced on hours a year, at about $5 an hour above 1,500 h and less as volume grows.

Ask for a quote

For people and teams

Captioners, editors and production teams who want proof to hand their clients. Payments are not open yet, so ask for access and we send a trial link on these terms.

Solo and Team compared

Solo

$29/mo

or $290 a year

Request access

Team

$149/mo

or $1,490 a year

Request access
Caption hours a month10 h40 h
Above that$3 an hour above 10 h, up to 30 h a month$3 an hour above 40 h, up to 120 h a month
People15 seats, then $20 a month each
A signed record and verify link for every file
Roles for owners, editors and viewers
CSV export of every file and answer
The API

Not in Solo or Team: the library report written for counsel or procurement, and the caption rows of an ACR, accuracy sampled by a person, invoice or purchase-order billing. Those come with the audit and Program. Records name who submitted each file.

Free, for anyone

  • Fix one file

    The converter, 5 runs a day, no account.

    Open the converter
  • Free account

    A library with 2 caption hours a month of signed records on your own files, and 10 converter runs a day. A hard limit: nothing is ever billed.

    Create a free account

The questions your counsel will ask.

Answered before they ask them.

What do you need from us?
Just the caption files, in SRT, VTT, TTML, SBV or ASS. Most video platforms can export them. You do not need to send video unless you want a person to check accuracy on a sample.
Does a passing record mean we are compliant?
No, and be wary of anyone who says otherwise. A record shows what was checked, against which version of which standard, on which date, and what the result was. Compliance is a legal conclusion, and it covers more than captions. The record is the evidence you bring to that conversation.
Can software really tell if captions are accurate?
Not from a caption file alone, and we will not pretend it can. Timing, placement, reading speed and format can be checked by the engine. Accuracy needs someone to compare the words against the audio, so on a paid audit a person samples it and the library report states the sample size and what it supports.
We already use a captioning vendor. Why would we need this?
Because you are currently taking their word for it. We do not write captions or sell fixes, so we can check a vendor's work, or your own team's, with no stake in the answer.
Who can verify a record?
Anyone you give it to. They paste it into captionpass.com/verify, and our server confirms it is exactly what was issued. That check runs on our server, because records are signed with a key only we hold.
What happens to our files?
They are used to produce your record and nothing else. They are not used to train anything. The privacy page lists what is kept and for how long.

See your own library, on the record.

A free account checks 2 caption hours a month of your own caption files, with a signed record for each one. No card.