Features

AI captions: word-by-word or phrases, one video or fifty

Vyra captions from its own transcript. Word-by-word or 3-5 word phrases, styled however you describe, and the same style across a batch.

By Sulan Zhang, co-founder of Vyra · Updated 2026-09-24

Vyra already has every word of your video timestamped. So captions are just a description. Word-by-word for TikTok, 3-5 word phrases if you want them readable, bigger on the words that matter, off during b-roll. Over MCP you can run the same style across fifty videos in one go.

How it works

  1. Upload a clip. Vyra transcribes it with a timestamp on every word.
  2. Describe the captions. Grouping, weight, color, position, which words to emphasize.
  3. Vyra puts caption items on the timeline. They're editable like any text.
  4. Fix a word by quoting it. "At 0:12 it should say 'gesso'."
  5. Reuse the style next time, or batch it from Claude, ChatGPT, or Cursor.

What you can ask for

You sayVyra does
"Word-by-word captions, TikTok style"One word at a time, timed to speech
"3-5 word phrases, white, lower third"Phrase groups below the face
"Make 'eighty percent' bigger and yellow"Emphasis on those words
"Copy the caption style from the reference, blocky, all caps, drop shadow"Matches it
"Turn captions off during the b-roll at the end"Time-ranged captions
"Keep captions off my face"Moves them

Prompts to paste

Add captions for all speech in this 45-second Reel. 3-5 word phrases, bold white text with a thin black outline, centered in the lower third. No animation.
Caption this word by word, bold, white, slightly above center so it clears the TikTok UI. Make "never again" and "eighty percent" bigger and yellow. No captions during the last 5 seconds of b-roll.
For every project in this folder, add 3-5 word captions in white with a black outline in the lower third, then export 1080x1920. Tell me when each one is done.

Compared to the usual way

Traditional editorAuto-caption buttonVyra
Getting text on screenType it or import an SRTOne click, one styleDescribe any style
Emphasis on a wordManual, per wordRarelyName the words
Matching a referenceEyeball itNoAttach it and say so
Fifty videosFifty sessionsFifty clicksOne instruction

Limits

  • Accuracy depends on the audio. Wind, music under speech, and crosstalk cause mistakes. Quote the fix and it's corrected.
  • Captions use the fonts in Vyra's catalog. The default is Inter Bold.
  • Translation isn't a caption feature yet.

FAQ

Can I edit a caption after? Yes. Click and type, or tell the agent which word to fix.

Word-by-word or phrases? Word-by-word for fast short-form. Phrases when people need to actually read it.

How do I caption a batch? Connect Claude, ChatGPT, or Cursor over MCP and give one instruction that loops over projects.

Related

  • /prompts/captions
  • /how-to/choose-captions
  • /features/footage-understanding
  • /mcp

Try it on your own footage

Upload your clips, say what you want, get the edit back. Free to start.

Start editing with Vyra