Preserve the instruction across sound, pictures, and text so more learners can follow the same demonstration.
By Arbab Naseebullah Kasi, Chief Executive Officer - CEO, Feel Worldwide Foundation Inc.
AI-generated illustration of educators reviewing a fictional training-video demonstration.
A trainer points to a corner of a cloth, folds it, and says, “Put this over here.” The recording is sharp. The captions reproduce every word. Yet someone who cannot see the movement still has to guess which corner moves, where it goes, and what the finished fold should look like. Nothing is wrong with the learner. The instruction has been divided between channels without checking whether each person can reach the whole explanation.
For educators and community teams producing short recorded demonstrations, the useful question is not simply whether a caption button exists. It is whether the essential instruction survives when a learner cannot hear the audio, see the picture clearly, or operate the player with a mouse. Captions, spoken descriptions, and a descriptive transcript do different jobs. Planning them together can make one lesson more coherent without turning it into three unrelated lessons.
Find the information that exists in only one place
Begin with the recording or storyboard, not a list of software features. Mark every moment where meaning depends on a gesture, an on-screen label, a change in position, a speaker's identity, or a sound. Then ask what the learner needs to know at that moment. A decorative plant does not explain a folding technique. The edge being aligned does. This distinction keeps the review focused on information rather than on describing everything in the room.
W3C's Web Accessibility Initiative separates the needs of audio-only, video-only, and video-with-audio content, as well as live and prerecorded media. Its media planning guidance is a useful starting point for choosing the appropriate alternatives. This article focuses on short prerecorded teaching videos. It is a practical production approach, not a complete accessibility audit or a determination of the laws that apply to a particular organization.
Write a brief information map for one difficult moment: what is heard, what is seen, and what someone must understand to continue. If the third answer depends on combining the first two, decide how to carry the missing meaning into the alternative format. Do this before spending time on transitions, music, or branding. A beautifully edited gap is still a gap.
Give captions a real accuracy review
Captions provide synchronized text for speech and the non-speech audio information needed to understand a video. They are not merely a summary of the narration. W3C also cautions that automatically generated captions are insufficient unless confirmed to be fully accurate. Use its captioning guidance to plan a human review, including meaningful sounds and speaker identification where needed.
In a fictional teaching example, two educators, Rosa and Daniel, record a demonstration of folding reusable event cloths and packing them into a tote. The first automatic caption changes “keep the patterned side outside” to “keep the patterned side inside.” Rosa checks the caption against both the recorded instruction and the intended procedure. Correcting spelling would not catch this problem. The review has to preserve meaning, timing, and the relationship between the words and the action.
They watch again with captions enabled on the actual page learners will use. A caption covers the cloth's lower edge just as Daniel aligns it. They change the framing or caption presentation so the relevant detail remains visible, then check the published player again. Their review note records the problem, the correction, and who checked it. A caption file marked “finished” is not enough if the final presentation obscures the lesson.
Put essential visual detail into the narration
For many instructional videos, important visual information can be included naturally in the speaker's script. W3C describes this as integrated description and distinguishes it from a separate described version or additional description track. Its guidance on describing visual information explains that the method depends on the content, available pauses, and player support. When the main audio already conveys the needed visual information, an additional description may be unnecessary.
Rosa and Daniel replace “Put this over here” with an original, more precise instruction: “Lay the cloth flat with the patterned side facing down. Bring the left edge across to meet the right edge, then smooth the fold.” They leave time for the movement rather than speaking over the next step. Later, Daniel explains that the packed cloths sit upright beside one another, so individual pieces can be removed without emptying the tote. The narration now communicates the action and its purpose.
They do not describe the table's grain, Rosa's clothing, or every hand movement. They describe what changes the learner's understanding. Nor do they assume that narration alone makes every physical activity suitable for every learner. Someone may need a different demonstration, adapted equipment, assistance, or another way to participate. Accessible information is an important part of access, not a substitute for listening to the person doing the task.
Make the transcript complete enough to stand alone
A basic transcript represents the meaningful audio. A descriptive transcript also includes the visual information needed to understand the video. W3C identifies descriptive transcripts as important for people who are both Deaf and blind, among other users. Its transcript guidance recommends making the text easy to find and organizing it usefully, rather than leaving it as a pile of caption fragments.
For the fictional cloth demonstration, the transcript begins with the purpose and materials, identifies the speakers, and follows the sequence of the recording. It preserves the spoken instructions and adds clearly marked descriptions where an essential visual detail remains unspoken. A brief introduction is labeled separately from the transcript so readers can distinguish editorial context from recorded speech. The final packing arrangement is included in words, not only in a screenshot.
A summary saying “This video explains how to fold and pack cloths” cannot do that job. Neither can a list of chapter titles. Rosa and Daniel ask a colleague to use only the descriptive transcript and explain the sequence back to them. The colleague notices that “repeat” does not say whether to repeat the last fold or the whole process for the next cloth. They clarify the recording and its alternatives together. Their test finds a concrete ambiguity; it does not certify universal accessibility.
Check the player, not just the files
A learner still needs to reach and control the media. W3C's media-player guidance includes keyboard operation, a visible focus indicator, clear control labels, and sufficient contrast. It also distinguishes these essentials from additional features such as playback speed controls and interactive transcripts. A well-prepared caption file cannot compensate for controls someone cannot find or operate.
Test the exact embedded player and page that learners will receive. Try starting, pausing, moving through the video, and enabling captions without a mouse. Check whether the transcript link is nearby and clearly named. Inspect a small screen and enlarged text. Record the device, browser, and setup used, because a successful check in one configuration does not establish that every configuration works. Where possible, include people who use the relevant assistive technology, with their time and contribution recognized.
Give learners a way to report a specific obstacle without disclosing a diagnosis publicly. “I cannot reach the caption control using my keyboard” is actionable information. Decide who receives that report, what temporary alternative can be offered, and who owns the correction. A contact route helps only when someone is responsible for following it through; do not advertise support capacity the team does not have.
Release one coherent lesson
Keep the video, captions, and descriptive transcript together as one versioned teaching resource. If a step changes in the recording, review the corresponding text and spoken description before replacing it. Give the team a simple release note containing the lesson title, version date, files checked, unresolved limitations, and reviewer. This is especially useful when one person records and another uploads: the final check belongs to the assembled lesson, not just to each person's contribution.
Start with one minute of an existing demonstration. Identify one instruction carried only by sound or only by the image, then revise it and its alternatives. Ask another person to tell you what they can now understand or do that was unclear before. This modest exercise turns accessibility from a label attached at the end into a production decision you can examine and improve.
Practical, inclusive learning connects with Feel Worldwide Foundation Inc.'s Education, Youth Development & Digital Access pathway. Educators or organizations interested in discussing an accessible-learning collaboration can connect with the foundation and describe the learning task and barrier they want to address. This is an invitation to begin a conversation, not a promise of a particular service or enrollment opportunity.
Sources reviewed October 14, 2026. Rosa, Daniel, and the cloth-packing demonstration are fictional teaching examples, not reports of FWF participants or activities.
