AI video clipping that runs on your own PC
Paste a link to a Twitch VOD, a Kick VOD or a YouTube video and hand it to Clippy the Cat. He watches the whole thing, picks the moments worth posting, crops them to fit a phone screen with the speaker kept in frame, burns in word-synced captions and writes the titles.
He does it all on your computer. Your footage is never uploaded, there is no subscription, and there is no cap on how many clips you make.
★ If you find it useful, give it a star. It costs nothing, and it is how developers find the app and end up improving it for you.
Clips Studio is free & open source ❤️
It runs entirely on your PC with no fees. Please consider donating to help cover development costs and keep it free for everyone.
For creators doing it all themselves
If you are growing a channel, you are probably the one filming it, streaming it, editing it and posting it. Clipping is how people find you, and it is usually the first thing that gets dropped. An editor is a cost most channels cannot justify yet, clipping tools charge for every video, and cutting them by hand takes a day you needed for something else. So the clips do not get made.
That is the reason this exists. Clips Studio runs on your own computer, so it costs nothing to use: no subscription, no credits, no charge per clip, no limit on how many you make. If you have a PC and something worth posting, you can promote your channel without paying anyone, me included.
Clipping tools like OpusClip charge a monthly fee or a price per video. Clips Studio does not charge anything. It runs on the computer you already own. Whether you are just starting out or already growing, the clips should not be the thing you have to pay for.
How it works
Every step runs locally. The only thing that touches the network in a normal run is downloading the video you asked for.
- Transcription faster-whisper with word-level timing, so captions land on the right syllable and clips start on a real sentence.
- Signal analysis Loudness spikes, laughter-shaped bursts, scene cuts, motion and on-screen reactions, measured across the whole video.
- Scoring A local language model reads the transcript alongside those signals, so a quiet line during an explosive moment is judged correctly.
- Framing Pose detection keeps the subject in frame, and in group footage the camera follows whoever is speaking. Crops are never stretched.
- Captions and titles Word-synced captions in your style, plus a title, description and hashtags, all editable before you export.
What you get
Finds what transcripts miss
Laughter, hype and clutch moments often read flat in text. Peaks in the audio and video become clip candidates in their own right.
Speaker-aware framing
In multi-person footage the crop follows whoever is talking, detected from mouth movement. Podcast mode frames each camera shot separately.
A real editor
Trim, cut, mute words, add a hook, music and a watermark, all non-destructive. Or describe the change in plain language and let the AI apply it.
19 languages
Translate, subtitle or dub a clip, with every line shown for review before anything is burned in. The app itself is translated into all 19 too.
Learns each creator
Recurring topics, running jokes and ongoing storylines build up per channel and sharpen the titles. It stays on your machine, and you can read or wipe it.
Long-form too
The same analysis produces horizontal clips, a best-of highlight reel, or the whole stream with the dead air removed.
What it's built for
Worth knowing before you download, so you're not guessing why a clip came out the way it did.
- People on camera, talking. IRL, just chatting, podcasts, vlogs, interviews. This is what the clip-picking and the framing are tuned for, and it's what gets tested on real streams before anything ships.
- Gaming and split-screen aren't shipped yet. Two things weren't good enough in testing: the framing didn't reliably find the part of the screen where the action was, and subject tracking mistook characters inside the game for the streamer. To a person detector, a person on screen is a person on screen. That produces clips centred on the wrong human, so it's held back rather than shipped half-working. It's being worked on.
- Reaction content isn't really its thing. Clips are chosen from what's said; it can't watch the video you're reacting to, so the moment that made the clip is invisible to it. Re-uploading someone else's footage is also a copyright question worth settling yourself.
Nothing here refuses to run. Feed it anything and it will produce clips. These are the cases where it's doing less thinking than it looks like, and where you should expect to reach for the editor more often.
Meet Clippy
Clipping a three-hour stream means watching a three-hour stream. Clippy the Cat does that part. He sits through the whole thing, notices where the room got loud and where the story turned, and comes back with the bits worth posting.
- He watches, you choose. Clippy brings back everything that cleared the bar, not a fixed three, so you pick what actually goes out.
- He remembers your creators. Running jokes and ongoing storylines build up per channel, so the titles get sharper over time.
- He works on your machine. Nothing is uploaded, nothing is metered, and he does not phone anyone about what he saw.
Why local matters
Nothing is uploaded
Unreleased footage, sponsor reads and personal streams never leave your computer. There is no account and no server holding your video.
No per-clip cost
Cloud clippers meter minutes or clips. Running the models yourself costs electricity, so processing a three-hour stream twice is free.
Yours to change
Nobody can take it away, put it behind a paywall, or change the terms on you. Even how it picks clips is yours to adjust.
Getting started
Download ClipsStudio-Web-Setup.exe and run it. A setup wizard takes it
from there. It checks your computer and downloads an AI model sized to your
graphics card.