How do I add captions to a video with AI?

Upload the video to Valmera and ask for captions, with the style in the same request: “white karaoke captions, size L, bottom of the frame.” It builds captions from a word-level transcript, burns them into the video, and renders a preview you check before exporting.

How accurate are AI captions?

Valmera transcribes with Deepgram nova-3, with a Whisper fallback, and times captions to each word. English is the best-tested language. Names, jargon and numbers are where errors show up: fix one by chatting (“at 1:32 it's Valmera”) and every caption updates.

Can I control how the captions look?

Yes. Pick a preset or set the color, highlight color, size (S to XL, with finer steps), position (bottom, middle, top or a custom height), animation, outline, shadow and backing panel, plus words to emphasize. Fonts come from a bundled set.

Why add captions at all?

Many viewers watch on mute, especially in social feeds, and captions make speech accessible to deaf and hard-of-hearing viewers and easier to follow for non-native speakers. They also help names and technical terms land.

Can I upload or download an SRT or VTT file?

No. Valmera creates captions from its own transcript and burns them into the MP4; there's no SRT or VTT import or export. If you need a separate caption file for a platform or player, create it from the final video with that platform's captioning or a subtitle tool.

Can I hide captions for part of a video?

Yes. Ask for captions to be turned off over a time range, for example while a title card or on-screen text is showing.

How much does captioning cost?

Creating an account and uploading are free. Editing, including captions, requires a subscription; see the current plans on the pricing page.

How to Add Captions to Any Video with AI

Upload the video, ask for captions in the style you want, check the preview, fix any misheard word by chatting, and export. With an AI video editor like Valmera, captions come from a word-level transcript, so they're timed to each word and burned into the final MP4. Here's the step-by-step, the style options and prompts you can paste.

Caption a Video Now

Upload it and describe the style. Creating an account and uploading are free.

Create an account →

See pricing →

How to add captions to a video with AI

1
Upload the video
Any video with speech: a Reel, a tutorial, a podcast clip, a product demo. MP4, MOV, M4V, MKV or WebM, up to 14 GB and 3 hours. Valmera transcribes it word by word when it indexes the upload.
2
Describe the caption style
Name a preset or describe the look: color, size, position, highlight color and animation. “Karaoke captions, white with a yellow active word, size L, bottom.”
3
Review the preview
The agent adds the captions, renders a preview and checks its own frames before reporting back. Read the captions against the audio, especially names and numbers.
4
Fix words and style by chatting
“At 1:32 it's Valmera, not Val Mira.” “Make them a bit bigger and move them up.” Corrections change the displayed words, never the audio.
5
Export
Export a full-resolution MP4 with the captions burned in. Ask for a 9:16 version in the same project if you need one; captions are laid out for the new frame.

Caption styles you can ask for

Presets

Clean, documentary, broadcast, podcast, reels, beast, karaoke, spotlight, elegant and more.

Word-by-word highlight

The spoken word lights up or pops as it's said: the short-form look.

Size and position

S to XL plus finer steps; bottom, middle or top, or a custom height.

Color and emphasis

Any text and highlight color, plus chosen words made bigger, boxed or accented.

Animation

Fade, pop, slide up, punch, bounce and other entrances.

Readability

Outline, shadow and a backing panel; bundled fonts only.

Multi-word presets sit at the bottom, clear of the face, by default; single-word styles such as spotlight can sit mid-frame. The full reference is in the captions docs.

Example requests to copy

Social captions

Add karaoke captions: bold white, size L, bottom of the frame, with each word turning yellow as it's spoken. Keep them clear of my face.

Clean tutorial captions

Add clean, readable captions in white on a dark backing panel, size M, bottom center, with a subtle fade. Nothing flashy, it's a tutorial.

Fix a word everywhere

The captions say “Val Mira”; it's “Valmera”. Fix it everywhere it appears.

Emphasis

Make the words “free” and “today” bigger and in the accent color every time they appear.

Hide captions in one section

Turn the captions off from 0:42 to 0:50 while the title card is on screen.

Got a word wrong? Fix it once

The transcript is the caption source, so you never re-time a subtitle file. Tell the agent the correct word, or edit it in the transcript panel, and every caption with that word updates, still timed to the audio. A correction changes what the caption says; it doesn't change what was spoken. The agent also audits the caption track for late, missing or overlapping captions when it reviews the preview.

Captions pair naturally with a reframe: ask for 9:16 and they're laid out again for the vertical frame. For long recordings, see how to turn long videos into clips.

Limits

Captions are burned in; there's no SRT or VTT import or export.
Captions cover the main recording; add text for inserted clips if they have speech.
No custom font uploads; English is the best-tested transcription language.

New to Valmera? Read how the AI video editor works, the getting-started guide, or caption videos from Claude or ChatGPT through the MCP server.