VideoPolisher
For demos, tutorials, talks, podcasts and website tours

Turn a rough recording into a polished, narrated video.

Record your screen once, talking as you go. VideoPolisher removes the dead time, rewrites and re-voices the narration, keeps the voice in step with what is on screen, and hands back a YouTube-ready video plus vertical shorts, captions, title and tags.

You approve every word of the script before anything is rendered. No re-recording, no timeline editing.

your recording
Ums, pauses, restarts, “where was I…”
polished
Clean narration, matched to the screen
16:9 demo 9:16 for Shorts & TikTok 30–60 s clips captions title & tags intro & outro
What you can make

Start from what you already have.

Pick a card. Every one says what you get back.

  • A rough screen recording becomes a polished, narrated video — as a demo, a tutorial or a talk, in your language or another, or with your video kept exactly as long as it is.
  • A finished video becomes short clips, a vertical version, or the same video in another language.
  • A script or an article becomes a podcast episode, as audio and as a video with slides.
  • A website becomes a narrated tour built from its own images and demo clips.
The problem

You recorded the demo. Now comes the part nobody enjoys.

The take is never clean

Long pauses while the page loads, the sentence you started twice, the click you had to redo. Cutting it by hand takes longer than recording it did.

Re-recording is expensive

A better script means a second take, a third take, and a room that is quiet enough. The screen part was fine. Only the voice needed work.

One video is never enough

YouTube wants 16:9. Shorts, TikTok and Reels want 9:16 with captions. LinkedIn wants a one-minute highlight. That is three more edits of the same thing.

Then the packaging

Subtitles, a title that people search for, a description, tags, a thumbnail, an intro and an outro. Each one small, together an afternoon.

How it works

Three steps. One of them is yours.

Pick what you want, see what will run, approve the script, download. No timeline, no keyframes.

1

Pick what you are making

Choose a card, answer a question or two, and add your recording. Tick the outputs you want; mark cuts, an intro or outro to keep, or a region to crop if you like. We show exactly what will run and what it costs before you commit.

2

Review the script

We transcribe what you said and rewrite it into clean narration that follows the order of events on screen. You read it, edit any line, and approve. Nothing is voiced or rendered until you do. Want to keep your own words? Approve as is.

3

Download or publish

A natural voice reads the approved script. The video is re-timed so the screen keeps pace with the narration, dead time disappears, and every format you picked renders. Download, or publish straight to YouTube.

Learn more

What each step does, and why it is there.

For when you want to know how it works before you try it. Every step is a separate piece of work: change one setting and only the steps that depend on it run again.

Clean up

Odd frame rates, wrong sizes, your cuts

Normalises the frame rate, scales or letterboxes to a standard size, crops to a region you draw, and removes dead time. Cuts are detected by AI, or you mark them on a timeline and they are used exactly as drawn.

Listen & look

Knowing what you said and what was on screen

Transcribes your voice with word timings and picks the key frames where the screen changed. Both feed the script and the sync.

Write the script

Rambling in, narration out

Rewrites the transcript into narration that follows the on-screen order, keeps your technical terms, and marks each sentence to the moment it belongs to. Keeps the spoken language or translates to another. You edit and approve — or take just the script and record it yourself.

Narrate & sync

A voice that stays in step with the screen

Natural voices from OpenAI or ElevenLabs, with pronunciation overrides for acronyms. The video is re-timed scene by scene so the narration lands on the action, speeding through dead stretches and never cutting a scene in half. Prefer the original length? Pauses are inserted instead.

Render

The file you actually upload

The polished 16:9 video, a transcript aligned to the final cut, SRT captions, YouTube metadata and a thumbnail. GPU encoding when available.

Vertical camera

Landscape apps on portrait phones

Reframes to 1080×1920 with a camera that follows the cursor or interprets the scene, stays inside the content, avoids black bars, and respects the safe zone under Shorts and TikTok overlays.

Pan & zoom

Widescreen that still reads

An AI-driven camera on the 16:9 output: zooms toward what the narration talks about, pans smoothly, snaps to edges, and holds still when nothing moves.

Subtitles & clips

Watched on mute, shared in pieces

Karaoke-style captions on the phrases that matter, in a font that fits the language. Then 30–60 second highlights with their own titles, for Shorts and for LinkedIn.

Intro, outro and inserts

Your brand, your existing clips

Mark an intro or outro to keep and it is passed through untouched and re-attached in every format. Or generate a branded intro and recap outro from the script. Splice other clips in at chosen times, loudness-matched.

Talking head

Your webcam bubble, handled

Detects and tracks the webcam bubble in a recording, covers the original with a card or logo so the camera can move freely, and places a clean still or your own headshot where you want it.

Podcast polish

The same engine, pointed at words

Start from a script, an article, or a link to one. It is rewritten for listening with chapters and show notes, fact-checked in a second pass, and narrated: an MP3 episode, plus video in 9:16, 1:1 and 16:9 with slides or images, captions and clips.

Website tour

Your site already has the pictures

Give it your site. It finds the core product pages and the demo clips and images already on them, writes one scene per feature, narrates it, and lays it out in 16:9 and 9:16 from a single script. Edit any scene's words and re-render only that scene.

Publish

The last click

Connect YouTube once and publish any output with its metadata. Titles, tags and descriptions are checked against YouTube's limits before upload.

Re-run only what changed

Tweaks should be cheap

Every output remembers what it was made from. Change a voice, a caption style or one line of script, and only the steps downstream of that change run again. Your approved script is never regenerated behind your back.

Who it is for

People who would rather ship the feature than edit the video about it.

Founders and indie makersLaunch videos and changelogs without a video person.
Product marketersEvery release in every format, on the day.
Developer advocatesTutorials where the voice keeps up with the terminal.
Course creatorsLessons recorded once, narrated cleanly, captioned.
Pricing

Pay for what runs.

Early access

Credits per minute of output. Start on the Free plan: polish recordings up to three minutes, with a watermark, no card needed. Plans add longer videos, every format and no watermark.

  • You see the cost of a run before you start it.
  • Re-runs after a tweak only charge for the steps that ran.
  • Plans and prices are published at launch. Early users keep their credits.
Request early access
Questions

Straight answers.

Do I need to re-record anything?

No. The screen recording is the source of truth. Only the narration is rewritten and re-voiced, and the video is re-timed to match it. If you would rather use your own voice, take the script and record it yourself.

Does it change what I said?

It rewrites for clarity and keeps the order of what happens on screen. It does not invent steps. You read the whole script and approve or edit every line before it is voiced.

Which languages?

Transcription detects the spoken language. Narration keeps that language by default or translates to another, with captions in a font that fits the script.

What can I upload?

Screen recordings in common formats such as MP4 and MOV, with your voice on the track. For podcasts, an audio file, a script, or an article URL. For a website tour, the site's address.

Whose voices are these?

Text-to-speech voices from OpenAI and ElevenLabs, chosen per project. Pronunciations of product names and acronyms can be pinned.

Where does my recording go?

Uploads are stored in our own cloud storage in the US and processed by AI providers to transcribe, write and narrate. Your files are yours, exportable and deletable on request. The privacy policy lists every provider we use.