clipsby kitmaple

Turn long videos into short clips — free, no sign-up, no watermark

Free AI that works online in your browser, without watermark. No upload: your video never leaves your device.

Clips runs on desktop Chrome or Edge. Phones and tablets are not supported yet. Open this page on your computer.

Honest limits

How it works

  1. Transcribe: drop an MP4 or MOV and press Transcribe. Whisper speech recognition runs in your browser and writes a transcript with a timestamp for every word.
  2. Pick lines: click the sentence where the clip should start, then shift+click or drag to the last one. Play the 9:16 preview, drag the frame sideways if the speaker is off-centre, and nudge the start or end by a quarter second.
  3. Export 9:16: the clip is encoded on your device as a 1080×1920 MP4 (H.264 video, AAC audio) with captions burned in word by word, then saved to the folder you choose.

Frequently asked questions

Is my video uploaded anywhere?

No. The video is read, transcribed, cropped, captioned and encoded in your browser, and the clip is saved straight to your disk. The only downloads are the speech model and its runtime; nothing you load is sent with them. The site counts anonymous events such as "clip exported" with coarse buckets like video length, never file names, transcripts or video data.

Is it really free, with no sign up and no watermark?

Yes. There is no account, no payment and no watermark or logo on the exported clip. Because the work happens on your own computer, there is no server time to pay for and no export limit beyond the 30-minute, 2 GB input size.

How does the AI transcription work, and which languages are supported?

The page runs OpenAI's Whisper base model (MIT licence) through transformers.js and ONNX Runtime Web, on your graphics card with WebGPU when available and otherwise on the CPU. You choose the spoken language from the list; English is the most accurate, and other languages work but make more mistakes with this small model. The transcript is kept in this browser, so reopening the same file does not transcribe it again.

How big is the download, and how long does it take?

The first transcription downloads the speech model from Hugging Face, about 80 MB on CPU or about 210 MB with WebGPU, plus a 27 MB runtime from this site. Both are cached by your browser for next time. With WebGPU a transcript is usually ready faster than the video plays; on CPU only it can take about as long as the video, or longer on older computers.

Which browsers and video files work?

Recent desktop Chrome or Edge. Phones, tablets, Safari and Firefox are not supported yet. MP4 or MOV files with H.264 video and AAC audio, up to 30 minutes and 2 GB, work best. HEVC (H.265), ProRes, VP9 or AV1 files may not open in every browser; convert them to H.264 MP4 first.

What does the exported clip look like?

A vertical 1080×1920 MP4 (9:16) with H.264 video and AAC audio, ready for YouTube Shorts, TikTok or Instagram Reels. The frame is cut from the centre of the picture unless you drag it sideways. Captions use one fixed style: bold white words with a dark outline, the spoken word highlighted in yellow.

Can I edit the captions or choose a caption style?

Not in this version. Captions come straight from the transcript, so a misheard word appears in the clip as it was heard. There is one caption style and no caption editor; if a line is wrong, pick a nearby segment or transcribe again with the correct language.

Does it find viral moments or follow faces automatically?

No. There is no automatic highlight detection and no face tracking; you choose the lines, which keeps the tool predictable and keeps everything on your device. You can export as many clips from one video as you like.