Video Kit: short videos with captions that move with your voice

Drop clips anywhere on the page, or press Choose clips (videos, photos, and sound files; MP4 and phone MOV files work best) or From Graphics Library. Editing starts the moment a clip is in: nothing waits for an upload. Each video becomes a piece on the timeline; a photo becomes a 4-second piece you can stretch to 30 seconds.

Behind the scenes, each clip from your computer is copied to the cloud in parts. The line under the timeline shows how far along each copy is (Copying to the cloud: talk.mp4 40%). If the connection drops or you close the tab, it continues from the parts still missing the next time Video Kit is open. Once a clip is in the cloud, a saved video opens on any computer or phone where you sign in. If cloud copies aren't switched on for your account yet, the line says Your footage stays on this computer: editing works the same, and the video reopens on the computer that has the files.

How long the cloud keeps footage depends on your plan: Free keeps the footage of your last 3 videos for 30 days; paid plans keep more. The line under the timeline shows your plan's numbers. Your finished videos aren't affected: they live in your Graphics Library.

Video Kit also keeps a copy of each clip in this browser, so a reload picks up right where you were (Pick up where you left off? Resume). If you open a saved video on a computer that doesn't have a clip and the cloud no longer has it, a yellow note names the clip with a Find button to point to the same file.

The first video sets the size: 9:16 (tall, for Reels, TikTok, and Shorts), or 1:1 when the footage is square. Change it any time under the video: 9:16, 1:1, 4:5 (feed post), or 16:9 (YouTube, a website). Video Kit makes videos up to 3 minutes; a longer edit shows a note with ways to shorten it.

The video plays exactly as it will export: captions, words on screen, cards, B-roll, the color look, and the sound mix. Play or pause with the round button or the Space bar (click the video first); the arrow keys step one frame (Shift for a second); the slider jumps anywhere.

The timeline shows every piece in play order, its width its length, with its sound drawn inside. Click a piece to select it and jump there; drag a piece to move it earlier or later; drag its left or right edge to trim it. Intro and Outro cards sit at the two ends; green marks show B-roll and orange marks show words on screen. Zoom (right of the buttons) makes the timeline wider for fine trims.

Split cuts the piece under the playhead in two (or press S). Remove piece takes out the selected piece. Undo and Redo (Cmd or Ctrl + Z, Shift for Redo) cover every change: buttons, drags, words, clean-ups, and plain-words requests. Add clips and From library add more at the end.

With a piece selected, a row under the timeline frames it: Fill the frame (covers the frame; drag the picture on the video to choose what stays in view) or Whole picture (all of it, with a soft blurred copy behind). Follow the face finds the speaker's face through the clip and keeps it in frame as they move, which matters when wide footage goes into a tall video. Face finding runs in your browser with open-source face detection; it reports its progress, and if it can't load or finds no face, you frame the clip by dragging instead. Zoom and Volume set this piece's zoom and loudness.

Make captions (in the Captions tab, or Make the words in the Words tab) listens to every clip and writes down each word with the moment it was said. Only the sound is sent for this, cut into short pieces of about a minute; the footage stays where it is. Captions then appear in time with the voice, a few words at a time, with the word being spoken lit up.

Style: Highlight the word (the spoken word changes color and grows a little), Word in a box (the spoken word sits in a colored box), Plain line, or Fill as spoken (words change color as they're said). Font (Motion Maker's 31 fonts), Letters (Soft shadow, Outline, Glow, 3D, Neon, Gradient, Retro offset), Size (S to XL), Words at a time (1 to 8), the words' color, the spoken word's color, a box behind the words (none, dark, light, or your brand color), ALL CAPITALS, and Where (Top, Middle, Lower, Bottom). Drag the captions on the video to put them anywhere. Show captions turns them off and on.

To fix a word's spelling in the captions, double-click it in the Words tab (or select it and press Fix spelling), type, and press Enter. The sound doesn't change.

If speech-to-text isn't switched on for your account yet, Make captions says so plainly, and you can still put words on screen yourself in the On screen tab.

The Words tab shows everything that was said, in order, with the word playing now highlighted. Click a word to jump there. To cut something, click its first word, shift-click its last word, and press Delete (or Cut in the bar that appears): the video cuts those words out, sharing the pause around them so the cut sounds natural. Cut words stay in the list, struck through; click one (or select several and press Bring back) to put it back.

Remove pauses and ums lists every long pause and every um or uh it found, each with a checkbox and the moment it happens (press the time to hear it). The top line says how many will be cut and about how many seconds that saves. Untick any you want to keep, then press Cut. One Undo puts them all back. Before captions are made, it finds quiet stretches from the sound alone (no ums yet).

Type or say any change in the Ask for any change box and press Make the change (or Cmd or Ctrl + Enter): "cut the part where I cough", "make it 30 seconds for Instagram", "put the price on screen when I say it", "bigger yellow captions at the top", "add an intro with our logo and upbeat music", "make it black and white". The examples under the box are shortcuts; anything can be asked.

One AI request reads what you asked against the edit (the pieces, every spoken word with its time, captions, words on screen, music, cards, and size) and turns it into the same steps the buttons make. It works on the edit, never by sending your video. The result lists anything Not done yet first, each with the reason and what you can do instead, then what was Done. Undo this change puts the edit back the way it was before. Thumbs up or down rates the result; thumbs down asks what went wrong, and Send and fix it asks again with your report.

What it can do: cut words or a stretch of time, keep only a stretch, make the video a length (pauses go first, then the end is trimmed), remove pauses or ums, fix a caption word, change the size, change any caption setting, add or remove words on screen at a moment, change or remove the music and its volume, add an intro or outro with a style and words, change framing or follow a face, reorder or mute pieces, show the logo, add B-roll from your Graphics Library at a word, add narration from Voice Kit, and set the color look. Things it can't do yet, such as speed changes, transitions between pieces, stickers, or translation, are listed under Not done yet, never skipped quietly.

On screen adds words that aren't captions, like a price, a name, or a call to action. Type them and press Add: they appear at the playhead for 3 seconds. Each one has its words, Starts and Shows for (seconds), a box (none, dark, light, brand, or a button), and a size. Drag it on the video to move it.

Brand: your brand color (from your Logo Kit brand kits, or any color), your logo (Pick a logo from the Graphics Library, or one from a brand kit), and In the corner the whole time with a corner. Color look sets the footage's look (Natural, Black & white, Warm, Cool, Vivid, Soft, Film) plus Brightness, Contrast, and Color; the words stay as they are.

Intro card and Outro card are title cards at the start and end, drawn with Motion Maker's animation styles: pick a Style, a headline, a second line (on the outro it becomes a button, like a web address or Book this week), what's behind it (your brand color, the video, black, or white), how many seconds, and whether the logo shows.

Pick one of the aiKit Originals (the play button beside each one lets you hear it) or Use my own music (MP3, M4A, or WAV, saved to your Graphics Library). Music dips under the voice by itself while someone talks and comes back up in the gaps (Dip the music while someone talks); set its volume and fade in or out. aiKit Originals are free to use anywhere, ads included.

Narration from Voice Kit: write a script, pick a voice, and press Make narration; the narration plays under the video from the start. Narration comes from Voice Kit; until Voice Kit is switched on for your account, this box says so.

B-roll is a picture or clip shown over the speaker for a moment while their voice keeps playing. Suggest B-roll reads what was said, looks through your Graphics Library for pictures that match words like a product or a place, and lists up to six with the word and the moment; press Add on the ones you like. Pick a picture for "word" adds one by hand at the word under the playhead. In the video, each shows for 2.5 seconds with a slow push; change the seconds or remove it in the list. B-roll starts at its word, so it moves with your cuts; if its word is cut, it doesn't show.

A Presentation Builder deck can become a video: open aikit.ai/tools/video-kit?fromDeck= followed by the deck's id (the Presentation Builder's Make a video button opens this). Video Kit reads the deck and builds a new edit with one piece per slide, drawn in the deck's own theme colors and fonts and laid out for the video's size: 16:9 to start, and still readable if you switch to 9:16, 1:1, or 4:5. Titles rise in, then each bullet follows; Big numbers slides show each number large; pictures in the deck show beside the words (above them on tall sizes).

The speaker notes become the narration: Voice Kit reads them aloud as one script (a slide without notes gets a short pause, long enough to read it), and each slide lasts exactly as long as its part of the narration. Each slide carries its own stretch of the voice, so cutting words, removing pauses, reordering slides, and music that dips under the voice all keep the picture and the voice together. Captions come from the script, so the words are exactly what was written; their timing is spread across each spoken line, so a word can light up a moment early or late.

While it works, the page says each step (Reading your presentation, Reading the speaker notes aloud, Laying out the slides). Anything it couldn't do is said at the top: no narration if Voice Kit isn't switched on for your account (the slides then get a reading length each), a sample sound if Voice Kit's real voices aren't set up yet, slides past Voice Kit's 6,000-character read that weren't narrated, or only the first 40 slides. Everything after that is the normal editor: change the size, add music or an intro, ask for any change, and export the MP4.

Name the video at the top. Save keeps the edit in your account (it then saves again by itself a few seconds after each change); My videos lists your saved videos with a picture, length, and date, opens one, starts a New video, or deletes one (after a confirm).

Export MP4 then Make the MP4 draws every frame in your browser at full size (1080 × 1920 for 9:16) with the sound mix, faster than the video plays on most computers, with a progress bar and the time left. Stop cancels. When it's done you can watch it, Download it, and Save to Graphics Library when the file is 25 MB or less (the library's limit per file); a bigger file is downloaded instead. Browsers without the fast encoder record the video in real time instead.