Captions, animations, camera moves and transitions are picked from what the speaker says. Then you fine-tune it like any video editor. Everything is visual, so it works with the sound off.
Vertical clips work best. Captions, animations, camera moves and transitions are added for you, then you can fine-tune anything on the timeline.
Your video stays in this browser and is never uploaded. What you say is transcribed right here in your browser, word by word. The first time, the speech model loads once (about 77 MB) and is reused after that. You can also add captions or paste a transcript.
Effects and music are generated live in your browser, so there are no files to license. Press play to hear them. Your video’s own sound stays off until you turn it on with the speaker button under the timeline.
Each line below has its start and end time. Fix any word, then press Re-plan the edit. Captions stay synced to your voice.
Added at the playhead for 2.6 seconds. It replaces that moment’s animation.
Templates are grouped by what they do. The Core set (marked in blue) are the ones that come up most in talking-head videos and are listed first in each group. Search by name or by what someone might say. Tap a color dot under any card to preview it in that color; in the studio every animation has the same color choice. Each card loops its animation and lists the words that trigger it. In the studio, any of these can be swapped onto a moment. Effects tagged “Edits your footage” transform your uploaded video itself: zooms, freezes, splits, grades and clones. Every other effect is a graphic layered over your video while it keeps playing.
Type something a speaker might say. The matcher ranks the animations that fit and plays the best one.
The animation starts on the timestamp of the trigger word and finishes within 1.8 s, so it never trails behind the speech.
At most one overlay plus captions on screen. A new overlay waits until the previous one has faded out.
Overlays stay in the top 40% of the frame. Captions sit under the chin and above the app's name and description area. The face box is never covered.
Entrances overshoot slightly and settle back. Exits are short fades. Nothing moves at a constant speed.
At most one word in five turns yellow: numbers, names and the trigger word.
No music or sound effects. Every animation has to make sense with the sound off.