Captions on every word.
Timed to the voice.
Captions drawn on your video from its own words, in the style you pick. Fix any word before you render.
- is0.02 s
- so0.16 s
- beautiful.0.48 s
- Someone1.34 s
- 4,5001.74 s
- years2.58 s
- ago2.94 s
- put3.26 s
- this3.50 s
- rock3.72 s
- right3.96 s
- here.4.18 s
- Correct.4.44 s
- This4.82 s
- pyramid4.98 s
- dates5.42 s
- back5.80 s
- to6.18 s
- about6.42 s
- 4,5006.62 s
- years7.58 s
- old.8.02 s
- The8.62 s
- passageway8.74 s
What does it do?
Captions let people follow a video with the sound off. The AI writes down every word with its timing, you pick a style and fix what it misheard, and the captioned video downloads as an MP4.
Every word, in your style.
- Every word, timed
- Speech recognition writes down every word with its timing. Clear audio comes out best, and any word can be fixed.
- Your style
- Choose a caption style, font and colour, and see it on your video before you download.
- In sync
- Each caption appears as its words are spoken, word by word, without timing anything by hand.
- Files too
- Download SRT, VTT or plain text, made from the same words.
- Right to left
- Most spoken languages. Arabic is captioned right to left, as it is read.
- Easy export
- Download the captioned video as an MP4, ready for TikTok, Reels, Shorts or anywhere else.
Timed like speech,
not like a file.
A caption that stays up after its word was said reads as a mistake. These are the rules the captions follow, with the numbers the app really uses.
Word. A word stays lit for as long as it could take to say: 700 ms, or 110 ms a letter for a long one, and never past the next word.
Page. A caption page leaves 600 ms after its last word when silence follows, not when the next page starts.
Speakers. A second speaker gets a colour only when two people each say at least 6 s of the words on screen. A diarizer that splits one voice in two would otherwise paint a monologue.
Files. SRT, VTT and text files are written in the browser from the same words, broken into phrases at a pause, a sentence end or a new speaker.
An aligner gave “doing?” the silence after it: 4.48 s lit for one word. The rule shows it for 0.70 s.
1 00:00:00,020 --> 00:00:01,060 is so beautiful. 2 00:00:01,340 --> 00:00:03,960 Someone 4,500 years ago put this rock
Questions
How does the auto caption generator work?
Upload a video or paste a link. The AI transcribes every word with its exact timing, and the captions are drawn on your video in the style you choose. You see the captioned preview before you download anything.
Can I fix a word the AI got wrong?
Yes. The video’s words sit beside the preview: click a word to correct it, and the captions change with it before you render.
Which languages can it caption?
Most spoken languages, including English, Spanish, Portuguese and Arabic. Right-to-left languages like Arabic are captioned right to left.
Does it work for vertical videos?
Yes. Captions are placed for the video’s own shape, vertical, square or wide. To turn a wide video vertical first, use Reframe or AI Clipping.
Is it free?
You can start on the free plan. Longer videos and more minutes come with Pro; the pricing page lists what each plan includes.
