Auto-emoji
Emoji captions that land on the right word.
One per line, on the word that earns it, timed to the moment you say it. Not an emoji on every noun.
LumaCaption adds at most one emoji per caption cue, attached to the meaning-bearing word in that line and popping in on that word’s own timing — matched from a curated keyword list rather than a broad dictionary, because broad dictionaries misfire far more often than they delight.
The emoji-caption look came out of short-form and it works, but the version most tools ship is the wrong one: match every word against a big dictionary and decorate whatever hits. What you get is four emoji in a seven-word line, half of them attached to words that were not the point, and a caption that reads as clutter.
The convention that actually works is one emoji per cue, on the word carrying the meaning, arriving exactly when that word is spoken. That is what this does, and the restraint is the feature.
One per cue, maximum
A line gets at most one emoji. The first meaning-bearing match in the cue wins and the rest are ignored.
Timed to its own word
The emoji pops on the timing of the word it belongs to, not at the start of the line — so it lands on the beat you said it.
A curated list, not a dictionary
Matching runs against a small hand-built stem list. Broad dictionaries produce more misfires than good hits, which is why this one is deliberately narrow.
Numbers become a chart
A bare number or percentage in a cue maps to a growth chart — the one case where the emoji is about the shape of the claim rather than the word.
What it matches, and why the list is short
The map covers the vocabulary that short-form video actually runs on: money and business, hype and energy, mind and learning, growth and goals, outcomes and failure, attention, feeling, time and scale. A few dozen stems in total.
The temptation is to make it bigger. The reason not to is that emoji matching fails asymmetrically — a correct emoji is a small pleasant beat, and a wrong one is a caption that looks automated, which is the exact impression the whole product exists to avoid. A narrow list that hits less often and misses almost never is the better trade.
- Money and business
- money, cash, rich, paid, price, free, sale, deal.
- Energy and reaction
- fire, hot, boom, crazy, insane, wow, shocked.
- Growth and outcome
- grow, launch, scale, viral, goal, focus, win, best, fail, mistake, stop.
- Numbers
- Any bare number or percentage in the cue maps to a chart, which is why stats lines get one automatically.
The thing nobody mentions about emoji in video
Emoji are drawn from the platform’s own emoji font, which means the same character looks different on an iPhone, an Android and a Windows machine — different art, different proportions, sometimes a different reading of the same emotion.
For a caption burned into a video this matters less than it sounds, because the export renders on the same device as the preview: what you see while you are styling is exactly what ships in the file. It does mean a teammate reviewing on a different platform may see slightly different art in the editor than you did, which is worth knowing before someone reports it as a bug.
How it works
Caption the clip
Word-level timings come from your audio. The video file is never uploaded.
Let the emoji land
One per cue, on the meaning-bearing word, timed to when you said it.
Style and export
Combine with any of 242 caption styles. Exports are free and unlimited on every plan, with no watermark.
Common questions
How many emoji does it add per line?
One at most. The first meaning-bearing word in the cue that matches the list gets the emoji, and everything after it is ignored. That is the convention short-form settled on, and it is what stops captions turning into clutter.
When does the emoji appear?
On the timing of the word it belongs to, not at the start of the caption. If the matched word is the fifth word of the line, the emoji arrives when you say the fifth word.
Why did my word not get an emoji?
Because matching runs against a small curated stem list rather than a broad dictionary. That is deliberate: a wrong emoji makes a caption look automated, which costs more than a missing one gains. Numbers and percentages are the exception — those always map to a chart.
Will the emoji look the same for everyone watching?
In the exported video, yes — emoji are drawn from the platform font at render time and the export happens on the same device as your preview, so the file carries exactly the art you saw. A teammate reviewing in the editor on a different operating system may see different emoji art than you do.
