Useful captions accurately represent speech and relevant non-speech audio, appear at the right time and remain readable without hiding important action. Review automatic captions yourself, then watch the exported video on the intended screen. Text on a timeline is only the beginning of the check.
Try a teaching example
An edit is a sequence of decisions.
An original illustrated exercise. The event and work are fictional.
Select a shot to see its job in a short paper-bookmark sequence.
A corner bookmark, made from one square.
Show the result: Begin with the finished object so the viewer knows what the steps will produce.
Your turn: make this with your own footage. Review whether each action is understandable and whether the actual speech, meaningful sounds and captions agree.
What captions need to communicate
Captions make audio information available as text. They can include speech, who is speaking when this is not otherwise clear, and meaningful sounds needed to understand the content. A decorative headline over a reel is not automatically a complete caption track.
A viewer may be deaf or hard of hearing, watching in a noisy place, or simply have the sound off. Good caption planning begins while you shape the edit, rather than as an export-day surprise.
Start with accurate words
Automatic transcription can save time, but it can misunderstand a name, a local expression or a quiet sentence. Compare the text against the actual audio. Correct factual details from the original source instead of guessing what sounds plausible.
For a bilingual video, ask someone fluent in the relevant languages to review spelling, meaning and natural phrasing. Translating and transcribing are different tasks; decide what your audience needs before producing either.
Example to practise
Fictional narration: “Fold the top corner towards the centre.” Check: does the caption match the actual instruction? Meaningful sound, if present: [timer rings] Check: would a viewer miss a necessary cue without that sound description? Include a sound description because it helps understanding, not as decoration.
Match the words to the action
Keep captions in step with the speech. Give the viewer time to read them and follow the picture. A sentence that flashes past before it can be read needs a different treatment, even if its spelling is perfect.
Break text at a sensible phrase boundary. Watch the whole sequence rather than judging one caption in isolation. The reading effort changes when the picture also asks someone to follow a quick action.
- Check the first and last word against the audio.
- Look for text that appears too early or stays after the topic changes.
- Review rapid cuts and busy frames with extra care.
- Let the content and audience guide reading time; one timing preset is not a universal rule.
Make space for text and picture
Use a readable size and sufficient contrast. A consistent backing panel can help when the background changes. Avoid covering the hands in a craft demonstration, the speaker’s face or other information needed to follow the story.
The publishing platform may place controls over the video. Check its current preview and guidance, and inspect the actual uploaded version when you can. One fixed “safe area” template cannot account for every platform and device.
Choose the format for the destination
Some destinations support a separate caption file that viewers can switch on or off; others use text rendered into the picture. Check the delivery requirements. Do not assume a subtitle file attached to your editing project will accompany an ordinary video export.
Keep an editable transcript or caption source so corrections are manageable. Test a separate file in the intended player. If text is rendered into the picture, review it again after compression and resizing.
Review the finished file twice
First watch without sound: can you follow the information the captions are meant to convey? Then watch with sound and compare accuracy and synchronisation. Neither pass replaces the other.
For the fictional bookmark exercise, record a short instruction, prepare its captions and ask another person to describe the step. Use their confusion to choose a specific revision. Your captions are helping someone make a bookmark, not auditioning for the role of smallest text on the internet.
- Words, names and numbers match the source.
- Relevant speaker changes and meaningful sounds are understandable.
- Text follows the audio and stays long enough to read.
- Captions remain legible without covering essential action.
- The exported file or caption track works in the intended player.
Official resources for further learning
The exercises and examples above were created for this guide. These references provide additional background; a link does not imply an endorsement or training partnership.
- W3C WAI: captions and subtitles (opens in a new tab)
- W3C WAI: understanding minimum text contrast (opens in a new tab)
- Blackmagic Design: official editing training (opens in a new tab)
Found an unclear explanation? Send Upnova a correction or question. Read our content and corrections approach.
Put the idea into practice.
Explore the video editing learning path, or ask how the current batch teaches and reviews this kind of work.