Skip to content
cliplimeClipLime

Add subtitles to a video automatically

Speech becomes captions on your own device. Edit, style, burn in or export SRT.

AI runs on your device, your audio is never uploaded

or drop it here, or paste from the clipboard

MP4, MOV, WebM, MKV, MP3, WAV, M4A · speech in 30+ languages

At a glance

What it does
Turns speech into timed captions with an AI model that runs on your own device, then lets you edit, style and export them.
Opens
Video: MP4, MOV, WebM, MKV. Audio: MP3, WAV, M4A, FLAC. Or import an existing SRT/VTT.
Saves
MP4 with captions burned in, SRT, WebVTT and a plain-text transcript.
Languages
30+ spoken languages, plus translation of non-English speech into English captions.
Styles
Sentence subtitles or short-form captions with word-by-word highlighting, five looks, custom size, position, colors and font.
Privacy
Your audio is transcribed locally. Only the open-source model is downloaded, once.
Price
Free, no watermark, no minutes limit.

How to use it

  1. 1

    Open a video or recording

    Drop the file in. Pick the spoken language, and whether you want captions in that language or translated to English.

  2. 2

    Generate

    The speech model downloads once and runs on your device. Captions appear live as each part is transcribed.

  3. 3

    Edit and style

    Fix any word, adjust timings, find and replace names, then choose a look: clean sentences or bold short-form captions.

  4. 4

    Export

    Burn the captions into an MP4, or download SRT/VTT files to upload to YouTube or your editor.

Speech recognition that never leaves your computer

Most caption generators upload your video to a server, transcribe it there and charge by the minute. ClipLime does the opposite. It downloads Whisper, an open-source speech-recognition model, into your browser and runs it locally, on your graphics card through WebGPU when it's available, otherwise on your processor. Your audio is never transmitted, which matters for client work, unreleased content and anything personal.

The model is downloaded once and cached by your browser, so the next video starts transcribing almost immediately. Choose Fast for quick drafts, Balanced for most videos, or Accurate when there are accents, jargon or background noise.

Sentence subtitles or short-form captions

  • Sentences, classic subtitles of up to two lines, split at natural pauses and punctuation. Best for YouTube, courses, interviews and anything watched with sound off on a big screen.
  • Short-form, one to six words at a time, big and bold, the style that keeps people watching on TikTok, Instagram Reels and YouTube Shorts. Pair it with word-by-word highlighting.

Because ClipLime keeps the timing of every word, you can switch between the two at any point and the captions are regrouped instantly, no need to transcribe again.

Make them look right

Start from one of five looks, Clean, Pop, Boxed, Lime and Karaoke, then adjust size, position, font, weight, colors, outline, shadow and background. Highlight the word being spoken colors or boxes each word as it's said, which makes fast speech easier to follow.

The preview is drawn with the same code that burns the captions into the export, so what you see on screen is exactly what you'll get. For vertical videos, check that captions stay clear of the app's buttons with the safe zone checker.

Fix mistakes quickly

No speech recognition is perfect, especially with names and specialist terms. Click any line to jump the video to it and type the correction. Times can be typed directly, lines can be merged or deleted, and Find & replace fixes a misheard name everywhere at once.

Already have subtitles? Import an SRT or VTT file to restyle them and burn them in, no transcription needed. To convert or re-time subtitle files, use the subtitle fixer.

Burned-in captions or a subtitle file?

  • Burn in (MP4) when the platform has no caption support or you want a specific look: Reels, TikTok, Shorts, ads, LinkedIn. The captions can't be switched off.
  • SRT or VTT for YouTube, Vimeo, course platforms and editing software. Viewers can turn them on and off, and search engines can read them.
  • Transcript for show notes, blog posts and quotes.

Tip: tighten the edit first with the silence remover, captions on a video without dead air read much better.

Questions people ask

Is my video uploaded to transcribe it?

No. The speech model runs inside your browser on your own device. Only the model itself is downloaded (from the Hugging Face CDN); your audio never leaves the tab.

How accurate are the subtitles?

Very good for clear speech in common languages, and better with the Accurate model. Expect to correct some names and technical words, the editor makes that quick, and Find & replace fixes repeated mistakes in one go.

Which languages are supported?

More than 30, including English, Spanish, Portuguese, French, German, Italian, Hindi, Urdu, Arabic, Turkish, Russian, Japanese, Korean and Chinese. You can also translate non-English speech into English captions.

Why does it download a model?

Speech recognition needs a trained neural network. Downloading it once to your browser is what lets ClipLime transcribe privately and for free. Browsers cache it, so it's only downloaded again if you clear your browser data.

How long does it take?

With a graphics card (WebGPU) most videos transcribe several times faster than real time. On a processor it's slower but still practical. The first run also includes the one-time model download.

Can I upload the SRT file to YouTube?

Yes. In YouTube Studio open the video, go to Subtitles, choose Upload file and select the SRT. WebVTT is accepted too.

Is there a length limit or a watermark?

No watermark and no per-minute limit. Very long recordings take more time and memory, so a desktop computer handles them best.