snqv finds the moments in a long video that work on their own, crops them to vertical with the frame following whoever is speaking, and burns in animated captions. Or start from just an idea and get a finished Short with a voiceover and footage.
There are two tools. Both run entirely in this page and share the same keys.
A podcast, interview, stream, lecture or vlog you want to cut into Shorts.
A topic, a fact, a story or a tip you want turned into a finished Short.
The parts that usually take an editor an afternoon, done in a few minutes.
The AI is told to start on a hook, end on a finished thought, and never overlap clips. Each one comes with a score and a one-line reason so you can judge quickly.
Face detection runs in your browser and smoothly pans a 9:16 window to the speaker. Slides or screen recordings? Switch to full frame over a blurred backdrop.
Words highlight as they're spoken. Choose Classic, Bold box or Minimal, then set size, position and highlight colour.
A live 9:16 preview plays with the exact crop and captions. Nudge the start and end by half a second until it's right.
Edit the script line by line, rewrite a single line with one click, and swap the footage for any scene. Nothing is locked in until you press make.
Upload as private to review first, or choose unlisted, public or scheduled. Titles, descriptions and tags are filled in for you.
Drop in the file. Paste its YouTube link too if it's public, so the AI can watch it without an upload wait.
Pick how many and how long. Read each clip's score and hook, preview it with captions, and trim.
Your browser renders each clip in 9:16. Download them all, or upload straight to YouTube.
Pick a style, length and voice. The AI drafts a title, description, tags and a scene-by-scene script.
Tweak any line, then click the footage you like for each scene, or search for something else.
The voice, footage and captions are assembled in your browser. Download it or upload it.
Behind the scenes: Gemini reads the video or writes the script, face detection and caption drawing happen on your computer, and the final video is recorded from a canvas by your browser.
snqv has no sign-up and no server. Instead you use your own free API keys, which stay in your browser. That's also why it costs nothing to use.
Finds clips, writes scripts and voices them. Google's free tier covers light use.
Get a free key ↗Free stock footage for new Shorts. Not needed for the clipper.
Get a free key ↗Only needed to upload to YouTube from the site. You can always download and upload by hand instead.
The Downloader tab saves a video you own, or have permission to use, so you can clip it here. A website can't download from YouTube on its own, so it talks to a small helper (yt-dlp) running on your PC; the video never passes through this site. The tab shows the one-time setup and whether the helper is connected.
https://snqv.lol as an authorized JavaScript origin.What it costs and what to expect, so there are no surprises.
| Part | Cost | Notes |
|---|---|---|
| snqv itself | Free | No account, no subscription, no watermark. |
| Gemini (AI) | Free tier, then pay-as-you-go | Video is analysed at low resolution to keep it cheap. Typically cents per video; very long videos cost more. Check Google's current pricing. |
| Pexels footage | Free | Subject to Pexels' licence and rate limits. |
| Rendering | Free | Done by your own computer, so nothing is uploaded for processing. |
Only clip, export or upload videos you own or have permission to use, and follow the terms of YouTube and the stock-footage provider. Face detection runs locally; no faces or video frames are sent anywhere.
No. If you paste a public YouTube link, Gemini watches it directly. If you only give a file, it's uploaded to Gemini's temporary file storage for analysis, using your key. Either way the clips are cropped, captioned and exported on your computer.
In this browser's local storage only. They're sent only to the service they belong to (Google, Anthropic, Pexels). Clear them any time on the API keys page.
Web pages aren't allowed to download YouTube videos, and snqv doesn't try to. The link lets the AI find the moments; to export them you provide the file, for example a video you uploaded yourself.
A small face-detection model loads in your browser the first time you press Play or Export on a clip set to "Follow speaker". It tracks the largest face and pans smoothly. If it can't load, the crop stays where the AI guessed the speaker is. You can also pick Fixed crop or Full frame.
Use the Trim controls to nudge the start and end by half a second, change the framing, or run the search again with a "Look for" hint such as "funny moments" or "practical advice".
Yes. The clipper offers lengths up to 90 seconds. YouTube allows Shorts up to 3 minutes, but shorter usually performs better.
You can browse and edit, but rendering is much more reliable on a desktop with Chrome or Edge.
Add a free key and try it on a 10-minute video.