Short-form video, rebuilt
Resculpt.ai rebuilds a short in your own voice.
Resculpt.ai is a web app for video creators. It takes a short video you own, rewrites its script sentence by sentence, speaks the new script in a voice you choose, covers the burned-in captions, and renders new ones synced to the new audio. Download the finished file, or schedule it straight to a YouTube channel you connect.
No editing timeline. One upload, five steps, a finished MP4.
Before
After
The hard part is the timing, not the words.
Swap a script for a new one and the audio stops matching the footage. A line that runs two seconds long pushes everything after it, and by the middle of the video the voice is describing a shot that has already gone. Resculpt.ai writes each sentence to the length of the one it replaces, then measures what the voice actually produced and repairs only the lines that missed — so the new audio lands on the original's cuts and b-roll.
How Resculpt.ai works
Five steps, each one re-runnable on its own. Change the script without re-transcribing, or re-voice without rewriting.
- 01
Upload
Drop in a vertical short you own, or paste a link to one.
- 02
Transcribe
Every word of the original narration, with its exact timing.
- 03
Rewrite
A new sentence for each old one, written to the same length.
- 04
Re-voice
Spoken in a voice you pick, then fitted line by line.
- 05
Render
New captions burned in, synced to the new audio.
What you get
Timed per sentence, not per video
Each new line is written to take as long to say as the one it replaces, so the audio lands on the original's cuts instead of drifting apart halfway through.
It measures, then repairs
After the voiceover is generated, any line that overran its slot is rewritten and re-voiced on its own. Nothing that already fits gets touched.
Your choice of voice
Pick from your ElevenLabs voices per project, and hear a preview before you commit to a full run.
Captions that match the voice
Timings come from the voiceover itself, not from re-transcribing it, so they cannot drift. Four styles, your own highlight colour, and a box that covers the original's burned-in text.
Music, ducked properly
A backing track that steps out of the way of the narration, using the voiceover's own word timings rather than guessing at an envelope.
Schedule it to YouTube
Connect a channel and pick a time. The title, description and tags are drafted from the script you actually ended up with, and you edit them before anything is queued.
Use footage you own
Resculpt.ai makes it quick to rebuild a video. That is not a licence to take someone else's. Re-voicing another creator's short and posting it as your own can breach copyright, and YouTube's reused content policy is enforced against an entire channel rather than a single upload. Bring your own footage, or footage you have cleared.
Questions
- What kind of video does this work on?
- A vertical short with a spoken voiceover. That is what the timing model is built around — it reads the narration's sentence boundaries and writes to them. Footage with no speech has nothing to rewrite.
- Do I need a YouTube channel?
- No. Publishing is optional, and every other part works without it. You can download the finished MP4 and post it wherever you like.
- Can I use someone else's video?
- You need to own the footage or have the rights to it. Re-voicing another creator's short and posting it as your own can breach copyright, and it falls under YouTube's reused content policy — which is enforced against the whole channel, not one video.
- Is the voice a clone of mine?
- Only if you have cloned it yourself. Resculpt.ai speaks your script in a voice you choose from your own ElevenLabs account, and declares the result to YouTube as synthetic content when it uploads.
- How long does a short take?
- A minute or two of processing for a sixty-second clip, most of it the voiceover and the render. Rendering happens on a server, so you can close the tab and come back.