Skip to content

Compliance evidence · Updated

The DCMP Captioning Key

The DCMP Captioning Key is guidance from the Described and Captioned Media Program for captioning educational media, not a law. Of the 4 elements CaptionPass grades, a caption file settles presentation rate and partly settles line count and division, speaker identification and non-speech information. Whether captions match what was said needs a person.

What is the DCMP Captioning Key?

A set of guidelines for captioning educational media, published by the Described and Captioned Media Program, which the U.S. Department of Education supports. The Key says it is a key for the vendors who caption for DCMP, and that it applies to anyone captioning media. Read the Key.

A media library may ask its captioning vendor to follow the Key. That makes it a contract term, not a legal duty.

Do we have to follow it?

Not by law. The Key states no legal requirement. The rules that bind public bodies name WCAG instead: ADA Title II names WCAG 2.1 Level AA, and Section 508 names WCAG 2.0 Level A and AA. The Key is useful beside them because it says what good captions look like in more detail than either.

What does the Key ask of a caption file?

CaptionPass grades 4 of its elements. Each is quoted below with how much of it a caption file can settle.

DCMP Captioning Key (educational media): what a caption file can settle
CriterionWhat it requiresWhat a caption file can settle
presentation-ratePresentation rateCaptions are presented at no more than 130 words per minute for lower-level educational media and 160 words per minute for upper-level.CheckedThe caption file is sufficient to settle this. A pass here is a real pass.Checked at the upper-level ceiling of 160 words per minute, per caption: the words on screen divided by the time the caption is shown. Lower-level media has a ceiling of 130, which this check does not apply, so a pass here does not show a lower-level file meets it.
line-treatmentLine count and divisionNo more than two lines per caption is preferred; three- or four-line captions are occasionally acceptable. A sentence broken across lines is broken at a logical point where speech normally pauses.Partly checkedThe file can show this failing, but cannot show it passing. No findings means nothing detectable is wrong — not that the criterion is met.A caption of more than 4 lines fails, checked on the file as delivered. Whether three- and four-line captions are only occasional is not checked, and neither is whether lines break at logical points. The Key sets no line length, so none is checked here.
speaker-identificationSpeaker identificationSpeaker changes are identified so the viewer can tell who is speaking.Partly checkedThe file can show this failing, but cannot show it passing. No findings means nothing detectable is wrong — not that the criterion is met.A half-applied convention is detectable. Whether an unlabelled cue needed a label depends on who was speaking, which needs the audio.
sound-effectsNon-speech informationSounds essential to understanding are described, in square brackets, distinct from dialogue.Partly checkedThe file can show this failing, but cannot show it passing. No findings means nothing detectable is wrong — not that the criterion is met.The convention is checkable wherever a description is present. Whether a sound was missed entirely needs the audio.

Cited from Described and Captioned Media Program, Captioning Key: Guidelines and Preferred Techniques (dcmp.org/learn/captioningkey, as retrieved on 2026-11-16).

What can a caption file alone settle?

Checked: the file is enough to decide these.

  • Presentation rate. Checked at the upper-level ceiling of 160 words per minute, per caption: the words on screen divided by the time the caption is shown. Lower-level media has a ceiling of 130, which this check does not apply, so a pass here does not show a lower-level file meets it.

Partly checked: the file can show these failing, but not passing.

  • Line count and division. A caption of more than 4 lines fails, checked on the file as delivered. Whether three- and four-line captions are only occasional is not checked, and neither is whether lines break at logical points. The Key sets no line length, so none is checked here.
  • Speaker identification. A half-applied convention is detectable. Whether an unlabelled cue needed a label depends on who was speaking, which needs the audio.
  • Non-speech information. The convention is checkable wherever a description is present. Whether a sound was missed entirely needs the audio.

What needs a person?

Everything that depends on what was said or shown. The Key’s first element of quality is accuracy: errorless captions are the goal, and it gives no percentage. Only a person comparing the captions with the video can judge that, and the same goes for a speaker nobody labelled or a sound nobody described. Caption accuracy requirements explains how that is measured.

Where does this come from?

How CaptionPass uses these grades on a library is on CaptionPass for compliance.

This page describes the rules as published and where they stood on 16 November 2026. It is not legal advice.

What this is not. A conformance receipt is evidence of what was checked, by which engine, against which version of which standard, and what it found. It is not a legal opinion or a safe harbour. The engine that issues it reads the caption file and never the audio, so anything that needs the audio — above all whether the captions match what was said — is reported as not evaluated rather than assumed. On the audit, Program and Contract, a person also samples accuracy, comparing clips with the video by hand, and the report states the sample size and what it supports. How accuracy is measured.