Long videos in.
Short clips out.
Choose a podcast, stream or talk. listens to it, finds the moments worth sharing, and makes vertical clips with word-by-word captions and framing that follows the speaker. Your video never leaves your computer.
How it works
Three steps, about as long as the video itself on an ordinary computer.
Choose a video
Your own video, or a campaign's source video that allows clipping. MP4, MOV, MKV or WebM. It is read straight from your disk.
It finds the moments
A speech model (Whisper) writes down every word with its timing, on your computer. Every sentence-to-sentence window is scored on its opening line, energy, completeness, topic words, pace and audience reactions, and you can see why each one scored as it did.
Download, post, track
Each clip is a 1080x1920 MP4 with captions burned in. Copy the caption and hashtags, post it, mark it posted, and type its views later to estimate what it earned.
Word-by-word captions
Bold white words, the one being said in yellow, placed clear of the app buttons and caption area.
Follows the speaker
A face detector steers a smooth 9:16 crop. Or keep the whole frame on a blurred background.
Even sound
Every clip is levelled to -14 LUFS, the loudness the big platforms play at, with peaks kept from clipping.
Private by design
No upload, no server, no account. Clips and notes stay in your browser; export a backup any time.
How earning by clipping works
Brands, apps, streamers and podcasters pay people to cut their long videos into short clips and post them. The Who pays page lists the programs we found, with their rates, rules, payout details, reputation and sources, plus the red flags to avoid.
- A campaign posts its rules. Usually: which source videos to use, clip length, required tags or links, and the platforms that count.
- You post clips from your own accounts on TikTok, YouTube Shorts, Instagram Reels or X, then submit the links.
- You are paid for counted views, most often per 1,000 views, sometimes per clip or from a prize pool. Views are checked, and clips that miss the brief can be rejected.
- Budgets run out. Many campaigns have a minimum number of views before a clip pays, a cap per clip, and a fixed total budget.
| Views counted | 50,000 |
| Rate (a common figure is about $1 per 1,000 views; programs list roughly $0.20 to $6) | $1.00 / 1K |
| Before any program fees | $50 |
Many clips get far fewer views than this, some get far more. Rates come from the programs' own pages and third-party write-ups listed on Who pays, and they change per campaign.
The honest part
- Earnings are not guaranteed. Most clips earn little or nothing; views decide everything, and programs can reject clips or hold payouts.
- Check each program's rules before you post: length, required tags and links, allowed platforms, payout countries and fees. Rules change per campaign.
- Only clip videos you have the right to use: your own, or source videos a campaign gives you permission to clip.
- Hashtags are keyword-based. They come from the clip's own words, not from what is trending. Edit them before posting.
- The moment finder reads words and loudness. It cannot see visual action, and it judges humour only by laughter it can hear. Watch each clip before you post it.
- Captions can mishear names and slang. Speech recognition is very good, not perfect.
Your video stays on your computer
Everything happens inside your browser tab: reading the video, listening to it, finding moments, drawing captions and encoding the clips. There is no server that could receive your video.
The one download
The first time you make clips, your browser downloads the speech model (77 to 510 MB, depending on the model you pick) from Hugging Face and keeps it. After that it works offline.
Nothing collected
No account, no cookies, no analytics, no ads. We cannot see your videos, clips, notes or earnings. Privacy details
Your data, your copy
Clips, campaigns and results live in this browser's storage. Download a backup file whenever you like, and restore it on another computer.
What you need
- A computer with Chrome or Edge, recent versions. Windows or Mac work best (they include the AAC sound encoder).
- A graphics card helps. With WebGPU the speech model runs partly on the graphics card. Without it, the processor does it all, a little slower.
- Phones and tablets: fine for browsing Who pays and tracking your posts. Making clips needs the power of a computer.
| Listening (balanced model) | about 4x real time |
| Rendering 1080x1920 clips | about 4x real time |
| A 110 s video to 3 finished clips | under a minute |
Questions
Is it free?
Yes. Because the work happens on your own computer, the site costs almost nothing to run, so there is nothing to charge for.
Can you see my videos?
No. The video is opened by your browser and never sent anywhere. The site has no server that could receive it.
Which languages work?
The speech model understands about a hundred languages and can detect the language by itself. Caption styling is made for left-to-right scripts.
How long can a video be?
Up to about three hours works on a computer with 8 GB of memory or more. Longer videos take more memory while it listens.
Does it post the clips for me?
Not yet. You download each clip and post it yourself; the studio keeps track of what you posted where, and the views.
Are you connected to the programs on Who pays?
No. The directory is independent research with sources for every figure. Always check a program's own page before you join.