How to check a video for typos before the client does

A typo on screen costs more than a typo in an email. It’s bigger, it repeats every time the video plays and the client spots it on the first watch, usually in the one place they know better than you: their own product name or their CEO’s title.

Editors miss them for a simple reason. You read what you meant to write, and by the fifth pass of a cut your brain fills in the word you expect. The way out is to check the text at the source, read the rendered frames on purpose and let software look at the pixels.

Know where the text in your video comes from

Spell check can only check text it can read as text. In a typical corporate or social video, on-screen text comes from several places, and most of them reach the timeline as pictures:

• Titles and lower thirds typed in the editor. These are text, so the editor’s spell check can see them.
• Motion graphics rendered in another app. A title animated in After Effects or Motion arrives as a video clip. To the editor, the words are pixels.
• Slides, charts and screenshots. A presentation exported as images, a product screenshot or a chart from the client’s deck.
• Text in the footage. A sign on set, a T-shirt, a phone screen or a screen recording of software.
• Captions and subtitles. Usually text, but often made by someone else, in another tool, late in the job.

Write this list for the video in front of you. Each source needs its own check, and the ones that arrive as pictures are where typos survive longest.

Approve the text before it reaches the timeline

The cheapest typo to fix is the one that never gets typed. Put every piece of on-screen text in the script or a separate text sheet, exactly as it will appear, and get the client to approve it there.

Run the document’s own spell check, then copy and paste from it into titles and graphics instead of retyping. Retyping is where “Managment” and swapped letters come in.

Names and job titles deserve their own list, confirmed in writing by the client. The lower thirds checklist covers that step in detail, including accents that some fonts can’t draw.

Use your editor’s spell check, and know its limits

Premiere Pro has had a real spell checker since the December 2021 release (22.1.1). The Graphics tab of the Text panel lists the text of every Premiere graphic and Motion Graphics template in the sequence, and spell check also covers captions and transcripts. Misspelled words get a red underline, and right-clicking shows suggestions.

Turn it on from the Text panel’s menu or from Edit > Spelling, and run it once the graphics are final.

It has two blind spots:

• It only checks text that Premiere owns. A rendered After Effects title, a slide, a screenshot or a sign in the footage is invisible to it.
• It isn’t always reliable. Users on Adobe’s forums reported the checker missing errors in some 2025 and 2026 versions. Check your version and your language setting before you trust a clean result.

Other editors and title tools vary, and many don’t check spelling at all. None of them can read text that’s already part of the picture.

Read the export as a separate pass

Read the file you’ll deliver, not the timeline. The export is what the client sees, after the graphics rendered, the fonts substituted and the last-minute fixes landed.

Make reading text its own pass, separate from watching for picture and sound. When you try to do both at once, you end up following the story instead of reading.

• Pause on every piece of text and read it word by word. Reading it backwards breaks the habit of filling in words.
• Read it out loud. Saying “manger” out loud sounds wrong in a way that reading it silently doesn’t.
• Check against the approved document, not against memory: names, titles, numbers, dates, URLs and prices.
• Use a fresh reader when you can. Someone who hasn’t seen the cut catches what the editor reads past.
• Repeat for every format. A vertical or square version reflows and re-crops text, so a word can break or get cut at the edge.

The final QC checklist puts this text pass in order with the rest of the last check before delivery.

Let software read the pixels

Tools that read the rendered video don’t care where the text came from. They pull frames from the export, run optical character recognition (OCR) on each one, check the words against a dictionary and give you a timecode for each suspect.

That covers exactly the sources the editor’s spell check can’t see: rendered titles, slides, screenshots and text in the footage.

Expect two kinds of noise:

• Stylized lettering gets misread. Outlined, condensed or animated fonts can turn a correct word into a false alarm. Each flag is a suspect to look at, not a verdict.
• Dictionaries don’t know your words. Names, brands and jargon look like typos to a dictionary. Good tools skip capitalized words or let you add a word list, which also means they won’t tell you a name is wrong.

A few options:

• Video Spell Checker is a Mac app built only for this job. You drop in a finished video, it reads the text and exports the typos as markers for Premiere Pro or DaVinci Resolve. As of October 2026 it’s in waitlist, with English dictionaries only.
• AI Review in RecReview runs on a version on demand, or on every upload with auto-review on. It reads the text on screen and flags clear spelling mistakes, names included, with the timecode, in English, Portuguese and Spanish. Each finding shows up in the comment list next to the client’s notes, and you can mark it resolved or dismiss it. On the next version, it also reads the text where the client asked for a fix on the previous one and marks each request done, partial or not done, or flags it as unclear or to check by eye.

No tool replaces the read against the approved document. Software catches the misspelled word. Only a person with the client’s list catches a name spelled right that belongs to someone else, last year’s job title or a price that changed.

A typo pass that fits into your QC

Run it in this order, because each step is cheaper than the next:

1. Before the edit: on-screen text approved in a document, names and titles confirmed in writing.
2. In the edit: text pasted from the document, the editor’s spell check run once the graphics are final.
3. On the export: an automated read of the frames, each flag checked against the picture.
4. Last: a human read of every piece of text against the approved document, once per delivery format.

Done this way, the client spends the review on the story and the pacing, not on spotting your typos.