Make a video useful without sound—or without seeing the screen

A viewer might be in a noisy train, using a screen reader, learning the language, or unable to hear a speaker. Another viewer may understand the narration but miss the crucial step shown silently on screen. A video that relies on only one route to meaning excludes people unnecessarily.

Accessibility works best as a production decision, not a caption button pressed after the edit. This guide follows a short instructional video from script to publication and gives a practical way to check whether its important information survives different viewing conditions.

Write down what the viewer must learn

List the essential facts or actions before writing the script. For a demonstration, that might be the tool being used, the setting selected and the visible result. Mark how each fact is communicated: speech, sound, picture or on-screen text. Anything carried by only one mode deserves deliberate attention.

Suppose the presenter says, “Click here, then change this.” The image may make the instruction obvious, but the words do not. “Open Export, then select the square format” carries the action in the narration. That small script change reduces the amount of additional description needed later and often improves the instruction for everybody.

Avoid forcing viewers to read dense text while listening to unrelated narration. Show one useful idea at a time and give it room. If a chart matters, explain the relationship it demonstrates rather than merely announcing that a chart is visible.

Distinguish captions from a transcript

Captions are synchronised text for the audio information a viewer needs, including speech and significant non-speech sounds. A transcript is a separate readable text version. W3C also describes a descriptive transcript, which includes relevant visual information. These formats complement each other: a downloadable transcript does not automatically replace captions in a video.

Use automatic transcription as a first draft. Listen through and correct names, specialist terms, numbers and negations. A missed “not” can reverse the meaning of an instruction. Identify a speaker when a change would otherwise be unclear. Include a relevant alarm or a door knock when it affects the story; do not bury the speech under labels for incidental sounds.

Edit captions for a person who has to read them

Break text at natural phrases. Avoid separating a person’s first and last name or placing a single word on a final line unnecessarily. Match the caption timing to the spoken phrase, while allowing enough time to read. Check fast exchanges carefully: a technically complete transcript can still be impossible to follow when shown too quickly.

Keep captions away from essential graphics. If the presenter points to controls at the bottom of the screen, reserve another clear area or move the graphics. Review the smallest expected display size; a caption that looks generous on an editing monitor may obscure half a phone screen.

Do not assume one caption file fits every edited version. A trimmed introduction shifts all later timings. Keep captions with the exact export they belong to, using the same version identifier described in our project handover workflow.

Make visual information available in words

Watch the video with your eyes closed, or ask a reviewer to listen without looking. Which actions become confusing? Integrate concise descriptions into the narration where possible. “The indicator turns green” conveys a result that “That looks right” does not.

For an existing edit, a separate described version or additional audio description may be appropriate. Choose the solution around the content and audience, rather than trying to squeeze narration into every available pause. A descriptive transcript can also explain diagrams, headings and demonstrations in a form that is easy to navigate with assistive technology.

Keep descriptions factual. Say what the image communicates and why it matters to the task. Avoid interpreting a person’s feelings from appearance alone or adding details that do not help the reader understand the content.

Publish a transcript that behaves like a web page

Use meaningful headings and short paragraphs. Link to the video and, where useful, include timestamps for major sections. Label who is speaking when the exchange would be unclear. Put instructions in a real list and data in a properly headed table instead of a screenshot of text.

A transcript can be lightly edited for readability without changing meaning. Remove transcription errors and distracting false starts, but do not silently invent clearer statements for the speaker. If an important factual correction is needed, correct the video or add a clearly identified content correction that applies to both versions.

Test the player, not just the files

Use the keyboard to reach the player, start and pause playback, adjust volume, enable captions and leave the controls. The focus indicator should remain visible. Check that controls have understandable labels and that fullscreen mode can be exited. Avoid autoplay with sound: viewers should decide when the experience begins.

Three practical reviews before publication

Review	What it exposes
Watch with sound muted	Missing captions, unexplained sound cues, unreadable timing
Listen without looking	Silent demonstrations and important unspoken text
Use keyboard only	Unreachable controls, invisible focus and keyboard traps

These are useful checks, not a claim of complete accessibility certification. When the audience or project requires it, include people who use relevant assistive technologies in the review. Record problems in ordinary terms—“caption button cannot be reached”—so they can be fixed and retested.

Keep the alternatives together

Place a transcript link near the player. Give language choices understandable names. If there is a separate described version, explain what it provides. Test the published page on a phone and at enlarged text sizes, including the transcript and its headings.

Clear source audio makes transcription and caption review easier. Our small-room recording guide starts with the room and microphone rather than a repair filter. Together, careful recording and deliberate alternatives make a multimedia page useful in many more situations.

Further guidance: W3C WAI: Making Audio and Video Media Accessible.