MAKING CLIPS

How to caption clips so everyone can read them

Updated September 28, 2026

SHORT ANSWER

How should I caption clips?

For short-form clips, burn captions into the video so they look the same on every platform and work with sound off, and also turn on the platform's own captions where it offers them. Keep lines short (the BBC's guide equals about 25 characters per line on vertical video), use high-contrast text, and keep captions clear of the app's buttons and bottom text. Check every name, number and key word by hand, because W3C guidance states that automatic captions are not sufficient unless confirmed accurate.

Burned-in captions vs platform captions

There are two kinds of captions. The W3C Web Accessibility Initiative defines them: most captions are closed captions that viewers can show or hide, while open captions "are always displayed and cannot be turned off" (W3C WAI). Burned-in clip captions are open captions.

	Burned-in (open) captions	Platform (closed) captions
How you make them	Your editor draws the text into the video frames	The platform generates them, or you upload or edit a caption track
Viewer control	Cannot be turned off, resized or restyled	Viewer can turn them on or off; some players let viewers set size and color
Look	Same on every platform; you control font, color and animation	Platform style
Where they are offered	Any editor	TikTok auto-generated captions and creator captions; YouTube automatic captions for videos and Shorts; Instagram automatic closed captions for Reels
Main risk	Errors are permanent once you export	Automatic text may be wrong; you must review it

Platform sources: TikTok accessibility help, YouTube automatic captioning and Instagram closed captions for reels.

For clips, use both when you can. Burned-in captions carry your style and work for sound-off viewers. Platform captions add a track that viewers can turn on and, in some players, restyle. Campaigns often ask for on-screen text: on 2026-09-25, half of the 22 public Clipster clipping campaigns asked for your own edit, text or captions (Clipster campaign data).

Readability rules with numbers

The most detailed public caption standard is the BBC's. It includes guidance for vertical 9:16 video. The table converts it to clip terms.

Setting	Guideline	Source
Line width	Up to 90% of the width of a vertical 9:16 video; up to 68% of a landscape 16:9 video	BBC Subtitle Guidelines
Characters per line	About 25 characters on vertical video (the BBC's equivalent of 37 characters on landscape)	BBC Subtitle Guidelines
Lines	Up to 3 lines on vertical video; up to 2 on landscape or square	BBC Subtitle Guidelines
Reading speed	160–180 words per minute; about 0.3 seconds per word minimum on screen (1.2 seconds for 4 words)	BBC Subtitle Guidelines
Size	Line height of 3.9–4.5% of the video height for 9:16 video. On a 1920-pixel-tall video that is about 75–86 pixels	BBC Subtitle Guidelines (pixel values are our calculation)
Contrast	At least 4.5:1 between text and background; 3:1 for large text	WCAG 2.2 success criterion 1.4.3
Color	White text on a black background is the BBC's default for legibility	BBC Subtitle Guidelines
Line breaks	Break at punctuation. Do not split an article from its noun ("the + table") or a preposition from its phrase ("on + the table")	BBC Subtitle Guidelines

Sources: BBC Subtitle Guidelines and WCAG 2.2 Understanding 1.4.3 Contrast (Minimum). WCAG treats open captions as images of text, so its contrast guidance applies to them.

Short-form styles often show one to four words at a time and highlight each word as it is spoken. These styles sit outside the BBC's rules, which are written for full-sentence subtitles. The trade-off is real: fewer words per screen reduce eye travel, but very fast word changes are harder to read. If you use a word-by-word style, keep each chunk on screen long enough to read, and keep phrases together.

Where to place captions: safe zones

Apps draw buttons, account names and post text on top of your video. Captions under that interface become unreadable. Platforms describe the problem in their ad guidance:

TikTok: "Do not place any text, logos, or key messages on areas where buttons, usernames, and captions may appear" (TikTok Ads Help).
Instagram and Facebook: for 9:16 Reels and Stories ads, "keep the edges (top, bottom and sides) free of key creative elements, text and logos". Meta's Ads Manager has a Safe zone guardrail that shades the covered areas (Meta Business Help).
WCAG: "Captions should not obscure or obstruct relevant information in the video" (W3C, Understanding 1.2.2).
BBC: define the caption area so that it does not cover the speaker's face, and check that a second line does not rise over the mouth.

A practical placement for vertical clips: put captions slightly below the center of the frame. That is above the bottom stack of post text and account names, and away from the button column on the right edge. Keep the speaker's eyes and mouth visible. Before you post, open a posted video in the target app, take a screenshot, and compare it with your frame.

Accuracy: names, numbers and meaning

Automatic captions are a draft, not a result. The W3C states that "automatically-generated captions do not meet user needs or accessibility requirements, unless they are confirmed to be fully accurate". Its example: the audio "Broil on high for 4 to 5 minutes. You should not preheat the oven" became "Broil on high for 45 minutes. You should know to preheat the oven" (W3C WAI). YouTube also warns that automatic captions "might misrepresent the spoken content due to mispronunciations, accents, dialects, or background noise" (YouTube Help).

Names. Check each person, brand and place name against an official source, such as the guest's own channel or website. The BBC advises against editing out names used to address people.
Numbers. Listen again to every number, price and date. A misheard "15" vs "50" changes the meaning.
Negatives. Check every "not", "can't" and "never". The W3C example above turns "not" into "know to".
Jargon and slang. Game terms, product names and memes are easy for speech recognition to mishear.
Verbatim quotes. Do not "improve" what the speaker said. Caption what they said, especially in debates and interviews.
Profanity. Follow the campaign's rules. Some want words masked; others ban certain words entirely.

Accessibility checklist

WCAG success criterion 1.2.2 requires captions for prerecorded audio in synchronized media at Level A. Its intent is that captions include not only dialogue but also speaker identification and "sound effects, music, laughter" needed to understand the content (W3C). For clips, that means:

Caption all speech, not just the key line.
Add important non-speech sounds in brackets, such as [laughter], [crowd cheering] or [music stops].
Identify the speaker when it is not obvious who is talking, for example when the camera shows the listener.
Keep captions in sync with the audio.
Do not use color alone to identify speakers. Add a name or a dash for a new speaker.
Turn on the platform's captions as well when available, so viewers can use their own settings.

A caption workflow that avoids rework

Transcribe automatically. Use your editor's or the platform's speech recognition as a first draft.
Correct before you style. Fix names, numbers and negatives first. Styling first means fixing text twice.
Split into readable lines. Break at natural points and keep within the line length above.
Style for contrast. Use a bold sans-serif font with an outline, shadow or box.
Place and check safe zones. Compare your frame with a screenshot of the target app.
Export and review on a phone. Watch the full clip on the device your viewers use, muted and unmuted.
Turn on platform captions after upload, and review them if the platform generated them.

Tools differ in how much of this they automate. FastClip, for example, burns in word-timed captions from the transcript when it exports a clip. Twitch clips can also include captions, which the streamer and channel editors can edit for 30 days after the clip is created (Twitch Help).

Questions

Should I burn in captions or use TikTok's auto captions?

Use both when you can. Burned-in captions look the same everywhere and work for sound-off viewers. TikTok's auto-generated captions add a track viewers can control. Review the auto captions, because TikTok lets you edit or remove them after posting.

What font size should captions be on vertical video?

The BBC recommends a line height of 3.9–4.5% of the video height for 9:16 video, which is about 75–86 pixels on a 1080x1920 frame. Many short-form captions are larger. Make sure the text still fits about 25 characters per line without covering faces.

How many words should be on screen at once?

For full-sentence captions, the BBC recommends up to 3 lines on vertical video at a reading speed of 160–180 words per minute. Word-by-word styles show fewer words at a time; keep each chunk on screen long enough to read.

Are automatic captions good enough?

No, not on their own. The W3C says automatic captions do not meet accessibility needs unless confirmed fully accurate, and YouTube tells creators to review and edit them. Check names, numbers and negatives by hand.

Where should captions go so the app does not cover them?

Keep them away from the right-side buttons and the bottom area with the account name and post text. Slightly below the center of the frame usually works. Check against a screenshot of the target app before you post.

Do captions need to include sounds?

For accessibility, yes, when a sound matters to understanding. WCAG's intent for captions includes sound effects, music, laughter and speaker identification.

Sources

We checked every claim on this page against these sources on the date above. Platform rules and rates change often. Found an error? Tell us and we will fix it.

FROM THE MAKERS OF THIS GUIDE

Cut talk clips faster with FastClip

Paste a YouTube or X link or upload a file, select the exact words in the transcript, and export a captioned clip. Best for podcasts, interviews and commentary. It does not crop gameplay to full screen or add watermark files yet. Free plan, no card.

Start clipping for free