Cut by what was said.
raea reads footage the way an editor does, by its words, and writes the edit as code it can render, check and change.
Both are the approved renders. The conversations are shortened; the code in Files is as raea wrote it.
- 01
Every word has a time.
raea times every word that's spoken. The Ides of March's narration became 485 words, each with its start time, and its cuts, captions and sounds are placed on those words.
content/rome/words.film.json
- 02
The edit is a file you can read.
Each piece is a React composition written with raea. The reel's stat card is set to land when she says “two percent”, so it follows her if the cut moves.
edit.tsx line 399: bigPop={cued("two percent")}
- 03
It hears the mix.
Music and sound effects are made for each piece and mixed under the voice. raea measures the result: The Ides of March came out at −14.2 LUFS against a −14 target.
measured with ffmpeg ebur128 on reference/approved.mp4
- 04
Approved means a person watched it.
Each reference piece was approved by a person, with sound, before it became the bar. New renders are compared with it frame by frame and on loudness.
case.json: ssim_min 0.95, lufs_tol 0.5 · reference/APPROVED
What happens after you ask.
The Ides of March, from its narration to the finished film, the way raea made it.
1It reads the narration.
raea times the narration word by word. The film's 4 and a half minutes became 485 timed words to cut and cue against.
the strike · 2:35.14

words.film.json · 485 words
Inside147.9Tillius148.9Cimber149.7came150.4forward150.7with151.1a151.2petition151.5and152.3seized152.7Caesar's153.3toga154.0Casca155.1struck155.9first156.5at157.1the157.3neck157.6Then158.6they159.1all159.4came159.7Caesar160.7fought161.5them162.0until162.3he162.8saw163.0Brutus163.3among164.1them164.7Then165.1he165.5pulled165.8his166.2toga166.5over167.2his167.5face167.9and168.5fell168.82It writes the film as code.
Every scene, caption and sound is in one React file. The toga tears on “seized”, the dagger is drawn on “Casca”, the body falls on “fell”: tied to the words, so if the narration moves, they move with it.
brand/rome/film.tsx · line 172
const FOLEY = [{ name: "cloth", at: W("strike", "seized"),src: fx("a heavy wool toga grabbed and pulled hard"), db: -14 },{ name: "draw", at: W("strike", "Casca") - 0.35,src: fx("a steel dagger drawn fast from its sheath"), db: -12 },{ name: "fall", at: W("strike", "fell."),src: fx("a heavy body collapsing onto marble"), db: -12 },{ name: "shatter", at: W("strike", "fell.") + 0.25,src: fx("a marble bust shattering on stone"), db: -13 },]W("strike", "seized") → 2:32.66 · "Casca" → 2:35.14 · "fell." → 2:48.82
3It renders every frame.
8,064 frames at 1920 × 1080, museum scans lit in 3D with fire and fog in the air, on the Mac's GPU in under ten minutes. Change a line and only the frames that line touches render again.
8,064 frames · 30 fps · 1920 × 1080












edit: Caesar's bust lit warmer · the strike's 2 of 12 shown frames render again
4It checks its own work.
Before you see anything, raea looks at every cut, checks each sound against its word, and measures the mix. This one came out at −14.2 LUFS, on target.
the mix · 4:29
−14.2 LUFS
- ✓lookevery cut and caption, frame by frame
- ✓checkeach sound on its word, the captions on the voice
- ✓measure−14.2 LUFS integrated, target −14
the strike · 2:35.14

words.film.json · 485 words
Questions
- Is raea an AI video editor for Mac?
- Yes. raea runs on your Mac: you describe the edit, and it transcribes your footage, writes the edit as code, renders it on your Mac's GPU and checks it before you see it.
- How does text-based editing work in raea?
- Every word of every take gets a start time. Cuts, captions and cards are pinned to words, so when the cut moves, everything tied to a word moves with it.
- Can it add captions automatically?
- Yes. Captions come from the word-by-word transcript and are checked against the voice: in the approved reels, each caption lands within a frame of the word.
- Does my footage leave my computer?
- Rendering happens on your Mac's GPU. The footage is read and rendered locally.
What would you make first?
Copy the prompt, paste it into your agent, and see what raea makes. It's free, and it runs on the computer you already have.
Paste this into your AI agent:
Install raea from https://raea.io/install and show me what it can make.