How the captions are made
The tool runs OpenAI's Whisper speech model (the "tiny" version, about 45 MB) inside your browser with WebAssembly. It listens to the soundtrack in 30-second windows and writes down each word with the moment it was spoken. The words are then grouped into short captions: a new caption starts when it reaches your word limit, at the end of a sentence or after a pause of more than 0.6 seconds.
Example: "Hello everyone and welcome back to the stream." at 3 words per caption becomes "Hello everyone and", "welcome back to", "the stream." In the highlight style, each word lights up in yellow as it's spoken, the look popular on TikTok and Shorts.
Accuracy and tips
The tiny model is fast and private, and accurate for clear speech in common languages, but it can mishear names, slang and words spoken over loud game audio. Every caption is editable, so fix those before you burn them in. Choosing the language instead of auto-detect helps with short clips. Lower-middle placement keeps captions clear of the buttons and descriptions on TikTok, Reels and Shorts; check with the safe zone checker.
The .srt and .vtt files work in YouTube Studio, Premiere Pro, DaVinci Resolve, CapCut and most players. Burning captions in re-encodes the video at its original size on your device. For the cleanest result, make the vertical version first with the vertical video maker, then caption it.
Frequently asked questions
Is this caption generator really free with no watermark?
Yes. There is no account, no watermark and no limit, because everything runs on your own device instead of a paid server.
Is my video uploaded to transcribe it?
No. The speech model is downloaded to your browser and runs there. Your video and audio never leave your device.
Which languages are supported?
Whisper understands about 100 languages; this page lists 18 common ones plus automatic detection. Accuracy is best for English, Spanish, Portuguese, French and German.
How do I add the captions in CapCut or Premiere?
Download the .srt file and import it as captions or subtitles. YouTube Studio also accepts .srt and .vtt under Subtitles.
Was this calculator helpful?