Captions
FrameCompose turns speech into timed caption blocks and keeps those blocks editable. Correct a word, move a line, change the look, or rebuild captions without flattening them into the video.
Maintained by the FrameCompose product team · Last reviewed 6 August 2026
Create captions from voiceover or transcribed speech.
Edit text and timing after generation.
Apply word-by-word and other visual caption styles.
Export the final captions as part of the rendered video.
Captions are created from the spoken track and placed against the project timing. Because they remain project data, a correction is a text edit rather than a new video generation.
That matters for names, product terms, accents, and technical language that automatic transcription can misunderstand.
Use readable placement, contrast, line length, and emphasis for the target format. Fast word-level captions can suit Shorts and Reels, while longer documentary lines may need calmer grouping.
The caption previewer helps compare looks before applying a style to the project.
Cutting a scene or changing narration can make captions feel early or late. Review the spoken line after structural edits and adjust the caption blocks where necessary.
The goal is not maximum animation. It is making the words easy to follow without hiding the subject.
Speech recognition can mishear names, punctuation, numbers, or overlapping voices. Always proofread captions, especially for branded, educational, medical, legal, or financial content.
A few quick answers before you jump in.
Yes. Caption text and timing remain editable after transcription or voiceover generation.
Yes. FrameCompose includes word-level caption styles as well as calmer grouped styles.
Yes. Styling captions is an editor action and does not require new scene generation.
No transcription system is perfect. Names, unusual vocabulary, noisy audio, and multiple speakers require review.
Explore the rest of the workflow, from first idea to final export.