Upload a video in Edit subtitles, review the Whisper draft against the recording, then choose a caption style and export. The original footage remains visible. Separate text accuracy, timing and visual checks before combining optional cuts, zoom and media.
See a real export
Punchy portrait captions
Pulse animation, four-word groups and a bright highlight over the original footage. · 66.0 s
Independent original human narration is combined with licensed film excerpts. The movie dialogue is not transcribed. Framing, captions and the declared audio/cut settings are applied. No synthetic voice or replacement performance is generated.
Watch the full example and see every setting ↗ · CC-BY-4.0
Source credits & licenses
(c) copyright 2008, Blender Foundation / www.bigbuckbunny.org
Original source · CC-BY-3.0
Mark Twain | Read by John Greenman / LibriVox
Original source · LibriVox-Public-Domain
See real examples
- Punchy portrait captions
- Cinematic neon subtitles
- Marker highlights on real footage
- Editorial reading: a quiet portrait
- Minimal landscape captions
- Typewriter narration in a square frame
- Outline captions: strong contrast
- Mono captions for a documentary look
- Bold portrait: two words at a time
- Wide neon story with a B-roll moment
- Mint marker: square social captions
- Editorial wide frame with clean audio
- Minimal portrait with image B-roll
- Typewriter: upper-case landscape
- Electric outline: square story
- Mono portrait with subtle music
- Caption-led cut: remove a sample pause
- A tighter opening: trimmed landscape
- A clean ending: square editorial cut
- Layered portrait: captions, music and B-roll
Choose the workflow that keeps your pictures
Edit subtitles accepts a local video with sound and retains its picture track. The music-video tool uses a soundtrack to create new background visuals, while text animation creates moving typography. Choose Edit subtitles when the source footage should appear in the result.
Correct the words before styling
Recognition is a draft. Listen to names, numbers, quiet passages and overlapping voices. Correct each caption and compare its start and end with the recording. Existing word timings come from the recognition model; replacing text uses estimated word divisions rather than a new speech-alignment pass.
Choose short caption groups and contrast
Compare pulse, marker, neon, editorial, minimal, typewriter, outline and mono styles. Start with a small word group and a position clear of important details. A long word or line may shrink to fit. Review the output on a phone-size display; more movement is not always more readable.
Review cuts and quiet-interval removal
Set trim boundaries and exclude unwanted caption rows. Quiet-interval removal detects low-volume audio, not meaning. It can shorten a useful pause or remove part of a quiet sentence. Check transitions after rendering, particularly around excluded rows and quiet speech.
Add zoom, a title and supporting media
Use rhythmic center zoom only when the crop preserves the subject. Add an opening hook, one scheduled image or video B-roll asset and a music track if they support the message. Keep music low enough to hear speech. Use media you have permission to publish.
Compare the public cases and reproduce a combination
Each public subtitle watch page lists its actual style, frame, timing and optional effects, with downloadable settings and captions. Film excerpts and independently recorded human narration are combined for these demonstrations; the narration is not dialogue from the movie. Source components and licenses are listed individually. No customer uploads are used.
Export and check the actual MP4
SRT contains text and timing but no font, colors, zoom or footage. MP4 contains the rendered visual treatment. Play the exported MP4 with sound from the opening through an edited section to the ending. Download within 24 hours. Inputs are limited to 100 MB and 10 minutes; Google sign-in and shared queue limits apply.
Common questions
Can I keep my original video and add subtitles?
Yes. Use Edit subtitles for captions over your original footage. Create video and animated text tools instead build new visuals from a soundtrack.
Can I download SRT without rendering a video?
Yes. After recognition and review, download the timed caption draft as SRT. MP4 export renders captions into the picture.
Can I change incorrectly recognized words?
Yes. Edit captions and line times before exporting. When a phrase is replaced, word timing inside it is estimated, so listen and check the exported result.
Does zoom follow a face?
No. The optional zoom is a rhythmic crop around the center of the frame. Check that important content remains visible.
Can I inspect real subtitle examples before signing in?
Yes. The public example gallery can be watched without signing in. Sources, caption files, licenses and settings are disclosed. Creating your own project requires Google sign-in.