09:07clean
9:16 · face tracking
How it works
The log of one job, start to finish
Below is the real timeline of a 13-minute talk. You paste the link; the rest happens in this order.
-
00:00you
Paste the link
A YouTube link or an mp4 file. Recordings up to three hours work. No sign-up needed: your first two clips are made as a guest.
https://www.youtube.com/watch?v=Nf7AoF-cTIYCreate Shorts -
00:09BeamTake
Only the audio is fetched
Not the video: only the audio is downloaded first. A 13-minute recording is 11 MB, so the first step takes seconds.
01 · Sourceaudio ready · 11 MB9 s02 · Transcript2/2 chunks · Turkish1 min 26 s03 · Clip selectionscoring candidates~30 s04 · Renderwaiting -
01:35BeamTake
The speech is transcribed word by word
Every word has a start and end time. Captions and clip boundaries both come from this, so no clip cuts mid-sentence.
-
02:10BeamTake
Candidates are found and scored
It picks segments that stand on their own and hook viewers in the first two seconds, and skips intros, sponsor reads and goodbyes. Each candidate gets a 0–100 score, and you can reorder them.
And after that, I started reading again01:03 · 52 s · face tracking5.9
Dear friends, for me you are the ones in the media who…08:41 · 38 s · face tracking5.4
Hello, I'm happy to be here today…00:04 · intro, skipped— -
03:40BeamTake
Framing, captions, render
In a horizontal recording the speaker is tracked to build a 9:16 frame, captions are burned in with the style you picked, and the clip renders in 1080p or 4K. Your brand kit's logo and template are applied automatically.
punch contrast 2 clips readytook 3 min 40 s -
lateryou
Download or schedule
Fix a misheard word in the captions, change the ratio, re-render. Connect your accounts and clips go out on their own at the times you choose, best-scored first.
Timings are from a 13-minute recording; a one-hour recording takes about 10 minutes. 9 caption styles, 9 ratios, 16 languages, up to 4K output.