Drop your footage
Pick the clips straight off your phone — one long take or twenty short ones. Nothing to trim first.
Drop in your phone footage — every take, every flub, all the talk in between. Deflub keeps the best take of each line, cuts the chatter, puts it in story order and adds captions. What comes back is ready to post.
Pick the clips straight off your phone — one long take or twenty short ones. Nothing to trim first.
Deflub listens to every word, picks the cleanest take of each line, drops the chatter and dead air, and lines the scenes up so the ad makes sense.
You get a vertical video with captions burned in, ready for TikTok, Reels and Shorts — 1080p, or your footage's full resolution up to 4K at 60 fps. Tweak a scene if you want — or don't.
Said it four times? It compares every delivery and keeps the one that's complete and clean — not just the last one.
AI reads every line and tells the ad from the talk between takes — "okay, again", mic checks, laughing at the product — and switches the talk off.
Your lines stay in the order you shot them (that's how you meant them), with a wordless hook moved up front when it finds one — and demos kept with the line they show.
Word-by-word captions with the spoken word highlighted, key words emphasised, and placed where TikTok's and Reels' buttons won't cover them.
Your footage is processed on your own device. Only when a device can't do a step does that one step go to our cloud — and the app tells you.
A native app for iPhone and Mac. Your projects stay where you made them.
If you'd rather be shooting the next video than editing this one, it's for you.
We read hundreds of reviews of the video editors people use today. These are the complaints that came up again and again, and our answer to each.
Silence removers guess from volume, so word endings get clipped and cuts land mid-syllable. Deflub times every word to the hundredth of a second and cuts only in the air between them — checked against hand-marked answer keys, not guessed.
Every cut lands on a whole frame, with a soft 25 ms fade so there's no click. Lip sync stays locked from the first cut to the last.
Every cut is on the timeline with the reason written on it — retake, filler, pause, check between takes. One tap puts it back. Every take of a line is listed; tap the one you like.
Captions come from the same word timings as the cuts and re-time themselves after every change. Tap a word to fix it; the captions follow.
No credits. Re-run it, undo it, save it again — none of it costs anything.
The price is on this page. Your first video is free and complete — no watermark, nothing held back.
It's a plain App Store subscription: cancel in Settings in two taps, and the App Store reminds you before it renews. No cancellation fee.
Your footage stays on your device. When an iPhone can't do a step, only the sound (and a few small pictures) is sent, processed and deleted — never kept, never used for training. You can switch that off.
Projects live on your device. Your footage is never changed or moved, and every edit can be undone or restored from a saved version.
No length limit and no file-size limit. Save in 1080p or at your footage's full resolution, up to 4K at 60 fps.
Your first video is free and complete. Prices in US dollars.
No. There's nothing to set up and no timeline to learn. Drop your clips in, and the finished video comes back. If you want to change something — swap a take, put a cut back, fix a word in the captions — it's one tap. The full timeline is there when you want it.
Talking-to-camera footage from a phone: product demos, hooks, UGC-style ads, tips. Several takes and chatter between them is exactly what it's built for. Music-only or no-speech footage isn't what it's for yet.
On a Mac, no: everything runs on the Mac. On an iPhone, the heaviest listening steps can run on our servers: only the sound (and a few small pictures for silent moments) is sent, processed and deleted — never kept, never used for training. You can switch that off in Settings; the video is identical either way.
iPhone and Mac (Apple silicon). Both are in early access — join the waitlist and we'll tell you when it's your turn.
On a recent Mac, about a minute for a few minutes of footage, and seconds to save the video. Slower machines take longer, or send the heavy step to the cloud.
English, for now. More languages come once each one is measured to the same accuracy.
Yes, any time: it's an App Store subscription, so you cancel in Settings in two taps and the App Store reminds you before it renews. No cancellation fee.
Some editors charge per AI action, so a redo or a mistake costs you. Deflub doesn't: re-run it, undo it, try another take, save it again — it's all included.
No. Every plan gets every feature, with no watermark.
We're letting people in a few at a time. Tell us what you'll use it on and we'll email you when it's your turn.
One email when it's your turn, the occasional update. No spam; unsubscribe any time.