Add captions to a video, free
Drop in the voiceover. Kinecard transcribes it in your browser, times it word by word, and puts the captions on screen — then turns the lines that carry the point into animated cards.
No account, no credits, no trial clock. Best on a desktop browser.
Four steps
- Drop in the voiceover and pictures. An mp3 and a folder of stills is enough.
- Transcribe. Runs in the tab, word by word with timings.
- Let it plan, then argue with it. Captions and cards are laid out for you; change any of them.
- Export. An mp4 lands in your downloads.
Captions, and then the thing captions cannot do
Word-timed captions are table stakes now. Everything does them, and a wall of moving text is still a wall of text.
What holds attention is when the screen draws what the line says: a figure counts up, two options sit side by side, a list ticks itself off as it is read. Kinecard reads the transcript, proposes which lines deserve that, and lets you overrule every one of them. Seven motion cards are free; Pro adds fifty-five more.
The captions you can actually change
Automatic is only useful if you can argue with it. Every one of these is yours after the transcript lands:
- The words. Names and jargon are what transcription gets wrong. Fix the word once and the timing stays.
- A word at a time, or a line at a time. One word landing on the beat reads as energy; a whole line reads as calm. Which one suits depends on the video, so it is a switch rather than a decision made for you.
- Where it sits, per caption. Height is set for the video and then overridden on any single caption — because the one time it matters is when a caption lands on top of something.
- How big it is. From barely there to filling the frame.
Portrait and landscape want different things
A Short is watched with a thumb hovering over it, so captions sit high enough to clear the interface and land a word at a time. A landscape video is watched leaning back, so a full line along the bottom is easier to read and less tiring.
Kinecard starts each orientation where it should start, and then gets out of the way. Nothing here is locked.
Nothing is uploaded
The speech model runs in your browser, so the audio never leaves your machine. There is no queue and no upload limit, because there is nothing to upload to. How that works.