1 Set Up

Understand captioning requirements

Why accessible captions matter

They are essential for many people who are Deaf or hard of hearing and can also improve access for people with cognitive and learning disabilities, including some individuals with attention-related conditions (such as ADHD) and other neurodivergent learners. Captions can also support viewers in noisy or quiet environments, people learning a language, and people who process information more effectively when they can read and listen at the same time.

Accessible captioning is more than producing a transcript of dialogue. Captions should represent the spoken words and important non-speech audio at the time that information occurs.

Core principle

A viewer who cannot hear the audio should be able to follow the video’s essential auditory information through the captions.

Understand captions, subtitles, transcripts, and audio descriptions

Captions

Captions are synchronized with the media and communicate spoken content—including lectures, classroom questions, relevant student comments or discussions—along with meaningful non-speech audio, such as a door slamming, applause, music, or an off-screen sound that contributes to understanding.

Subtitles

Subtitles commonly communicate or translate spoken dialogue. They may not include the non-speech audio information needed for accessibility, so subtitles should not automatically be treated as equivalent to accessible captions.

Transcripts

A transcript provides a text-based version of the meaningful audio content from a video or audio recording. It includes spoken information and, when relevant, identifies speakers and describes important non-speech sounds.

When are transcripts required? WCAG 2.2 AA requires a transcript or equivalent text alternative for prerecorded audio-only content, such as a podcast and prerecorded video-only (silent) content. For a prerecorded video with audio, captions are required, but a separate transcript is generally not required for Level AA. Providing a transcript can still be a helpful accessibility practice.

Audio Description

Audio description provides spoken descriptions of important visual information that is not already communicated through the video’s audio, such as actions, demonstrations, charts, on-screen text, or changes in a scene. It primarily supports people who are blind or have low vision by providing access to visual information they might otherwise miss.

For prerecorded video, WCAG 2.2 Level AA requires audio description when important visual information is not already provided through the existing audio track. In some cases, a separate audio-description track may not be necessary if important visual information is clearly described within the existing narration or dialogue.

Remember

Captions, transcripts, and audio descriptions serve different purposes. Captions provide access to audio, transcripts provide a written version of audio content, and audio description provides access to important visual information. Depending on the media, more than one may be needed to provide accessible content.

Determine what needs to be captioned

Prerecorded video with meaningful spoken or non-speech audio should be captioned. Live video with meaningful audio also requires an accessible real-time solution when captions are required for the event or content.

  • Caption lectures, training videos, tutorials, interviews, demonstrations, announcements, and other instructional or public-facing video with meaningful audio.
  • Ensure embedded or linked third-party videos have accurate captions when the audio provides information needed to understand the content.
  • For audio-only material, provide an accurate transcript and consider captions when the playback environment supports them.
  • If a video has no meaningful audio, captions may not be necessary, but important visual information may require another accessibility treatment.

Plan early

Check caption availability before selecting third-party media. It is much easier to choose an accessible video than to remediate a long video immediately before it is needed.

2 Organize

Prepare accurate caption content

Caption all meaningful speech

The caption text should accurately represent the spoken content. Do not routinely summarize, simplify, or omit speech just to make captions shorter. Preserve the speaker’s meaning, terminology, names, and important verbal information.

  • Include dialogue, narration, announcements, and relevant off-screen speech.
  • Use correct spelling, punctuation, capitalization, and terminology.
  • Verify names, course terms, technical vocabulary, acronyms, numbers, and proper nouns.
  • Do not clean up a speaker’s meaning in a way that changes what was actually communicated.
  • Minor speech disfluencies do not always need to be represented unless they are meaningful to the content, characterization, or context (e.g. class remarks unrelated to course content).

Include meaningful non-speech audio

Some sounds communicate information that a listener receives even when no one is speaking. Include those sounds when they contribute to the meaning, tone, action, or understanding of the video.

Too vague or missing More informative
[sound] [door slams]
[music] [upbeat music]
[noise] [audience applauds]
Nothing shown while an alarm sounds [fire alarm beeping]

Brackets are commonly used for sound descriptions. Describe the sound objectively and concisely. Background sounds that do not contribute meaningful information generally do not need to be captioned.

Identify speakers when the speaker is unclear

When multiple people speak, the viewer should be able to determine who is talking. Speaker identification is especially important when the speaker is off-screen, when voices overlap, or when a change in speaker is not visually obvious.

  • Use a speaker’s name when the person is known and the name is relevant.
  • Use a meaningful role when the name is not known, such as INSTRUCTOR or NARRATOR.
  • Keep the identification method consistent throughout the video.
  • Do not add a label when the active speaker is already unmistakable from the video and no confusion is likely.
Ambiguous Clear
Can you submit it Friday? INSTRUCTOR: Can you submit it Friday?
Yes, I can. STUDENT: Yes, I can.

Accuracy matters

Automatic speech recognition can create a useful draft, but the final captions should be reviewed by a person who can correct errors and understand the context.

3 Add Content

Build readable captions

Synchronize captions with the audio

Captions should appear when the corresponding audio is heard and remain on screen long enough to be read. Poor timing can reveal information too early, delay a response, or force viewers to choose between reading and watching the action.

  • Begin a caption close to the start of the associated speech or sound.
  • Remove it when that information is no longer being heard.
  • Avoid displaying a speaker’s response before the person says it.
  • Give viewers enough time to read without leaving outdated captions on screen.
  • Review fast dialogue and rapid speaker changes carefully.

Break caption lines at logical points

Line breaks affect readability. When a caption must wrap, divide the text where a reader would naturally pause rather than separating words that belong together.

  • Keep phrases and grammatical units together when possible.
  • Avoid separating an article from its noun, a preposition from its phrase, or a person’s first name from the last name.
  • Do not create a new caption solely because the text reaches a visually convenient point.
  • Keep captions compact enough that they do not dominate the video.
Awkward break More natural break
The accessibility team will
review the report tomorrow.
The accessibility team
will review the report tomorrow.
Dr. Maria
Santos will present.
Dr. Maria Santos
will present.

Guidance

The Described and Captioned Media Program (DCMP)/industry guidance recommends synchronized captions, generally one to three lines on screen, with captions visible long enough to read and positioned so they do not obscure essential visuals.

Position captions without hiding important visuals

Captions must be readable without covering information the viewer needs to see. This is especially important in tutorials, demonstrations, presentations, and videos containing lower-third names, charts, controls, or on-screen text.

  • Use the platform’s standard caption area when it does not obscure essential content.
  • Move or reposition captions when supported if important visual information appears in the same area.
  • Do not place critical labels, instructions, or graphics where captions are expected to appear.
  • Check the video on smaller screens as well as desktop displays.

Use readable caption presentation

Caption text should be easy to distinguish from the video. When the player allows users to control caption appearance, preserve those controls. For open captions that are permanently burned into the video, use a clear typeface, adequate size, and strong contrast.

  • Use mixed upper- and lowercase text rather than all caps for ordinary dialogue.
  • Maintain sufficient contrast between caption text and the video background.
    WCAG 2.2 AA requires a 4.5:1 ratio for normal-size text color to its background.
  • Use a background, outline, or other treatment when needed to keep captions legible as the scene changes.
  • Avoid decorative fonts and visual effects that reduce readability.
  • Do not make captions so small that they become difficult to read on common viewing devices.

Design for captions

When producing new video, reserve visual space for captions from the beginning rather than trying to fit them around a completed layout.

4 Format

Choose the right caption delivery method

Closed captions vs. open captions

Closed captions

Closed captions can be turned on or off by the viewer. When supported by the player, they can also allow users to adjust caption appearance. Closed captions are generally preferable when the delivery platform supports them reliably.

Open captions

Open captions are permanently visible as part of the video image. They can be useful when a platform does not support caption tracks, but viewers cannot turn them off or customize their appearance.

Choose intentionally

Select a caption method that remains available in the actual platform and format being distributed. Test the final published version, not only the editing environment.

Provide transcripts when appropriate

A transcript is a written version of the spoken content and other important audio in a video or audio recording. Transcripts provide another way to access and review the content and can support people with hearing, visual, cognitive, or learning disabilities. Unlike captions, transcripts are provided separately and are not timed with the video or audio.

A transcript should:

  • Include all meaningful speech.
  • Identify speakers when needed for understanding.
  • Include important non-speech audio, such as music, alarms, applause, or other relevant sounds.
  • Be organized so the content is easy to read and navigate.
  • Be updated when the video or audio content changes.
  • When are transcripts required? For prerecorded audio-only content, such as a podcast, WCAG requires a text alternative that provides equivalent information.
  • For prerecorded video with audio, captions are required, but a separate transcript is generally not required for Level AA.

Remember

Captions and transcripts serve different purposes. Captions are synchronized with the video or audio as it plays, while transcripts provide the content separately in writing. A transcript does not replace required captions for video with meaningful audio.

Captions do not replace audio description

Captions provide access to audio information, while audio description provides access to important visual information that is not conveyed through the audio. This may include actions, demonstrations, charts, on-screen text, or other meaningful visual details.

For prerecorded video, WCAG 2.2 Level AA requires audio description when important visual information is not already conveyed through the existing audio.

  • Review the video once without looking at the screen.
  • Identify essential information that is available only visually.
  • In new productions, incorporate important visual information naturally into the narration when possible.
  • For existing video, provide audio description or another equivalent method when needed.

Separate needs

Captions address access to audio; audio description addresses access to essential visual information. Some videos need both.

5 Check

Review and test before publishing

Review automatic captions before publishing

Platforms such as YouTube, Zoom and other video services can generate captions automatically. Automated captions can save time, but speech recognition may misinterpret accents, technical terms, names, background noise, punctuation, and speaker changes.

  • Treat automatic captions as a draft.
  • Watch the entire video while reviewing the captions.
  • Correct words, punctuation, capitalization, timing, speaker labels, and meaningful sound descriptions.
  • Pay extra attention to names, acronyms, specialized vocabulary, equations, dates, and numbers.
  • Review captions again after edits are saved or republished.

Final Review Before Publishing

Before publishing a captioned video, watch it with captions enabled and verify:

  • All meaningful speech is represented accurately.
  • Names, terminology, numbers, acronyms, and technical vocabulary are correct.
  • Meaningful non-speech sounds are included.
  • Speakers are identified when needed.
  • Captions appear at the appropriate time and remain readable.
  • Line breaks occur at logical points.
  • Captions do not cover essential visual information.
  • Spelling, punctuation, and capitalization are correct.
  • Automatic captions have been manually reviewed.
  • The final published or embedded version actually displays the caption track.
  • A transcript is available when it is needed as an additional alternative.
  • Essential visual information has been addressed separately when audio description is needed.

Do not publish unchecked captions

An automatically generated caption track is not reliable simply because captions are present. Accuracy and synchronization must be verified.

6 Export

Publish captions and verify delivery

Caption prerecorded video: recommended workflow

  1. Obtain the final or near-final video before captioning so edits do not invalidate the timing.
  2. Create a caption draft from a transcript, captioning tool, automatic speech recognition, or professional captioning service.
  3. Correct the spoken text and add meaningful non-speech audio.
  4. Identify speakers where needed.
  5. Adjust timing and line breaks so captions follow the audio naturally.
  6. Upload or attach the caption file to the video platform.
  7. Watch the published video from beginning to end with captions turned on.
  8. Correct any remaining errors and perform a final quality check.

Efficiency

When producing a series of videos, maintain a terminology list containing names, departments, course terms, acronyms, and other words that automatic captioning frequently misrecognizes.

Caption live video and events

Live captioning must communicate speech as the event occurs. Live captions are required under WCAG 2.2 AA. Furthermore, higher-stakes events or accommodation needs may require a qualified real-time captioner or CART service.

  • Identify the captioning method before the event.
  • Provide professional captioners with names, agendas, presentation materials, and specialized terminology in advance when possible.
  • Test audio quality, microphones, internet connectivity, and caption display before participants arrive.
  • Make sure remote and in-person participants can see the captions.
  • If a live session is recorded for later use, review and correct the captions before publishing the recording.

Live is different

Real-time captions prioritize immediate access. A recording that will remain available should receive a post-event caption review for accuracy and readability.

Adding captions to video in PowerPoint

PowerPoint supports captions and subtitles for media, but the exact workflow and supported caption formats vary by PowerPoint version and operating system. Current Microsoft 365 versions commonly support WebVTT caption files, and some versions also support UTF-8 SRT files.

  • Create and review the caption file for the video.
  • Select the video in PowerPoint.
  • Use Playback > Insert Captions, or the equivalent caption command available in your version.
  • Choose the caption file and insert it.
  • Play the presentation and verify that the captions appear, are synchronized, and remain readable.

Export caution

Microsoft notes that some media optimization and presentation-to-video workflows may not preserve separate captions or alternative audio tracks. Test the final distributed format.

Captioning video hosted online

For video hosted on a streaming platform, use the platform’s caption tools or upload a supported caption file. The exact steps vary by service, but the quality-assurance process should remain consistent.

  • Confirm that you have permission or ownership needed to add or edit captions.
  • Upload an accurate caption file or generate a draft using the platform’s tools.
  • Edit automatic captions rather than publishing them without review.
  • Confirm that captions can be turned on in the public or course-facing version of the video.
  • Test playback using the same link, embed, LMS, or website that the audience will use.

Common Accessibility Mistakes

Common problem Better approach
Automatic captions are published without review. Use automatic captions as a draft, then correct the complete video.
Only spoken dialogue is captioned. Include meaningful non-speech audio such as alarms, music, laughter, or applause when relevant.
Captions appear several seconds after the speech. Retiming captions so they closely follow the corresponding audio.
The viewer cannot tell who is speaking. Identify speakers when the visual context does not make the speaker clear.
Captions cover a demonstration or lower-third label. Reposition captions or redesign the visual layout to preserve both.
Long captions fill much of the screen. Break captions into readable units at natural language boundaries.
A transcript is provided instead of captions. Provide synchronized captions for video and use the transcript as an additional resource.
The video has captions but key visual information is silent. Provide audio description or another accessible equivalent for the visual information.
Captions work in the editing tool but disappear after export. Test and remediate the actual file or platform that will be distributed.

Quick Checklist Before Sharing

  • All meaningful speech is represented accurately.
  • Names, terminology, numbers, acronyms, and technical vocabulary are correct.
  • Meaningful non-speech sounds are included.
  • Speakers are identified when needed.
  • Captions appear at the appropriate time and remain readable.
  • Line breaks occur at logical points.
  • Captions do not cover essential visual information.
  • Spelling, punctuation, and capitalization are correct.
  • Automatic captions have been manually reviewed.
  • The final published or embedded version actually displays the caption track.
  • A transcript is available when it is needed as an additional alternative.
  • Essential visual information has been addressed separately when audio description is needed.

References and Further Guidance