How to Add Animated Captions to an Uploaded Video with AI

You can turn an uploaded video into a captioned social clip with Vibe Motion AI. The agent transcribes the source audio, uses word timestamps to place animated captions, and builds a composition around the footage. Your original video remains the source; the animation adds layout, titles, emphasis, and transitions.

What does the AI inspect?

The transcription tool inspects the audio track, not the pictures. It returns speech text and source-relative word timestamps. It does not detect visual highlights, read on-screen text, or recover subtitles burned into a video. If a silent demonstration needs labels, provide the wording and time ranges yourself.

Supported video sources for transcription are MP4, WebM, and MOV with an audio track, up to 100 MB and 10 minutes. The file is sent to ElevenLabs for transcription, which has a separate credit cost. Review the agent's disclosure before proceeding. Reusing the saved transcription avoids paying to transcribe the same unchanged source again.

1. Upload your video and describe the intended result

Open Vibe Motion, select Upload references, and choose the source video. Mention the uploaded file in your request so the agent knows which asset to use. A concrete starting prompt is:

Use my uploaded interview video. Make a vertical 9:16 social clip.
Keep the original footage and speech. Transcribe the spoken audio and
add captions synchronized to the words. Tell me the separate transcription
cost before proceeding.

Use large, high-contrast English captions below the footage, with no more
than two lines at once. Highlight important words in yellow as they are
spoken. Keep faces visible and preserve the footage's proportions.
Do not translate, paraphrase the speaker, or invent additional claims.
Add a short opening title based on the actual topic. No background music.

If the source is in another language, say which language the captions should use. Correct names and technical terms after reviewing the transcript. Word timestamps provide alignment, but they do not guarantee a perfect transcription.

2. Choose short sections of a longer source

You can use several time ranges from the same uploaded file. Give the source timestamps explicitly and distinguish them from positions in the finished animation. The following ranges are an example for a longer source, not the 25.4-second demonstration above:

From my uploaded recording, use these source ranges in this order:
00:30-00:36, 01:15-01:22, and 02:40-02:46.

Join them into a vertical highlight video. Keep each section's original
speech, and synchronize captions to the retained words after each cut.
Use a simple transition between sections. Do not extend the source ranges
or claim that these excerpts form one uninterrupted statement.

Those ranges contain 19 seconds of source material before any transitions or added holds. Vibe Motion places trimmed sections on the composition timeline; it does not create three smaller source files. Local export may still need to load the full original video. For a source larger than the upload or transcription limit, prepare a shorter source file first.

3. Revise captions for reading, not just decoration

Watch once with sound to check synchronization, then once muted to check comprehension. Look for names, punctuation, cut-off words, and phrases that disappear too quickly. Keep captions clear of faces and interface controls, and avoid animating every word so aggressively that it becomes hard to follow.

Keep the footage and timing. Correct the product name to "Kiwix"
where the transcript misspelled it. Use shorter caption groups, keep each
group to two lines, and reduce the bounce. Preserve the speaker's wording.
Do not move captions over the face or the bottom call to action.

The correction above is an example: supply the actual spelling from your recording. If the source already contains subtitles, choose a layout that avoids overlapping them. You may need to edit the original video separately if those embedded subtitles must be removed.

4. Keep speech audible and export the finished clip

Use the original speech once. If you add music, ask for a low level beneath the voice and a fade at the end. Listen for doubled speech, abrupt cuts, and background music masking quiet words. The caption animation and audio should still match after any clip rearrangement.

Open Export MP4, choose a resolution and frame rate, and select local or cloud rendering. For a 9:16 composition, 1080p gives a 1080 by 1920 output. Review the downloaded MP4 from start to finish: check that the footage moves, captions stay aligned, audio is present, and the last word is not cut off.

Are animated captions the same as an SRT subtitle file?

No. These captions are part of the rendered picture. An MP4 with animated text does not automatically provide a separate selectable subtitle track or SRT download. Use a separate subtitle workflow when the publishing platform needs a caption file. Keep the transcript and timing report as references.

For text-only animations, see kinetic typography. For a product story built from a webpage and screenshots, read the website-to-launch-video guide.