Generate subtitles for video automatically, free

Turn speech into subtitles with an AI model that runs in your browser. Edit the text, download SRT or VTT, or burn the subtitles into an MP4. No watermark, no sign-up, and your video and audio never leave your device.

  • Free
  • No watermark
  • No sign-up
  • Stays on your device

Choose the language if automatic detection gets it wrong.

AI model

Working…

0%

✓ Done

Before
After
Download

🔒 Your file stays on this device. Nothing is uploaded.

Automatic subtitles turn speech into timed text. Most online tools do it by uploading your file to a server, so your voice and footage end up on somebody else's computer. Clipalio runs Whisper, an open speech recognition model from OpenAI, inside your browser. Only the model is downloaded (about 45, 80 or 250 MB depending on the size), and only after you pick a file and press the button. It is then kept in your browser, so the next video starts at once.

You get a list of lines with start and end times that you can correct. Download it as SRT or VTT for YouTube, Vimeo or a video editor, or burn it into the picture so it shows everywhere, including TikTok, Instagram and WhatsApp. Whisper knows about a hundred languages and detects the spoken one itself. English and Spanish came out clean in our tests and Russian good; languages with less training data are weaker, so read the text before you publish.

How to use it

  1. Add your video

    Drop a file on the box or choose one. Nothing heavy loads until you do, and you see its length and a time estimate at once.

  2. Pick language and model, press the button

    Leave the language on automatic or choose it. The AI model is downloaded once, with a progress bar. Then your own device listens to the audio.

  3. Fix the text and export

    Play the video, correct words and times in the list, then download SRT or VTT, or burn the subtitles into an MP4.

Why use Clipalio

Your video stays with you

Most subtitle sites upload your file. Here video and audio never leave your computer; only the AI model is downloaded.

Free, no watermark, no minutes quota

No logo, no account, no monthly limit. Your own computer does the work.

Editable, with a live preview

See the subtitles over the video and fix any line or time before you export.

Honest about speed

You get a time estimate for your device before anything is downloaded, and a smaller model is suggested for long videos.

Frequently asked questions

Is my video uploaded to a server?

No. The audio is decoded and recognised on your device. The browser downloads the AI model (from the Hugging Face CDN) and a small runtime library (from jsDelivr); nothing about your file is sent back.

Which model should I choose?

Balanced (Base) is the default: on our English and Spanish test clips it was almost word-perfect. Fast (Tiny) is quicker but makes more mistakes. Accurate (Small) is best for Russian, accents and noise, but needs a strong computer and much more memory.

How long does it take?

On an Apple M-series Mac the Balanced model ran about 6.7 times faster than the video plays with the graphics card and 4.4 times with the processor only: roughly 1.5 to 2.5 minutes for 10 minutes of video. Older computers are slower, and the Accurate model on a processor is several times slower. The page shows an estimate for your device before you download anything.

What is the difference between SRT/VTT and burned-in subtitles?

SRT and VTT are small text files that players show over the video; viewers can switch them off and you can edit them later. Burned-in subtitles are drawn into the picture, show everywhere and cannot be removed, but the video is re-encoded.

Honest limits

  • Long videos take long, and a computer without a graphics card can need more than the length of the video. Use the Fast model or cut the video first.
  • Weak computers and phones may run out of memory, especially with the Accurate model.
  • Accuracy drops with music, noise, strong accents and languages with less training data (such as Arabic, Hindi, Thai or Japanese). That is why the list is editable.
  • Burning subtitles in needs WebCodecs (recent Chrome, Edge or Safari) and re-encodes the video. SRT and VTT work everywhere.
  • There are no speaker names, word-by-word highlighting or translation yet. The first run downloads 45 to 490 MB.