How to read a script on camera without it showing

There are two ways to film yourself saying a text, and both are bad.

Learn it by heart, and spend half an hour on forty seconds of video, delivered in the voice of a pupil reciting. Or read notes propped up beside the lens, and produce a take where your eyes come off the camera on every sentence. Everyone sees it, even without being able to say why.

Here is what actually fixes it, in the order that matters.

1. Read a script on camera at lens level, not beside it

That is the basic mechanical rule. The human eye is noticeable the moment it drifts a few degrees off the camera axis. Text lying on the desk fifty centimetres below the lens means a permanently lowered gaze.

The simplest solution for a video shot on a phone: scroll the text on that same phone’s screen, right under the lens. The offset becomes two or three centimetres, which is invisible on camera.

For a shoot on a proper camera, a mirror teleprompter does the same job: a half-silvered plate at forty-five degrees in front of the lens, with the text displayed below it and flipped so it reads correctly in the reflection.

2. Write for the voice, not for the page

This is the step almost everyone skips, and the one that makes the biggest difference.

A well-written text and a text that says itself out loud are not the same text. The first has subordinate clauses, asides, and words nobody ever pronounces. The second has short sentences and places to breathe.

Three simple corrections that transform any script:

• Cut every sentence over twenty words in two. No exceptions. You can glue one or two back together on the reread.
• Swap written connectors for spoken ones. “Nevertheless” becomes “except that”. “Furthermore” becomes “and then”.
• Write the breaths into the text. A blank line, a marker, whatever you like. A written pause is a pause you actually take.

The test that never lies: read the text aloud once. Every place you stumble is a place to rewrite, not a place to enunciate harder.

3. Set the speed to your real speaking rate

This is the number one cause of ruined takes, and it is a silly one: the text runs away from the second sentence, you speed up to catch it, and your voice goes mechanical.

A comfortable reading rate sits between 120 and 150 words per minute for most people. But your own number isn’t an average, it is data to be measured: read a passage whose word count you know for thirty seconds, and double it.

Once you know that number, everything becomes predictable. A 200 word script at 123 words per minute takes one minute thirty-eight. So you know before you shoot whether your text fits the format.

4. The framing changes with the format

A Reel doesn’t read like a speech, and a speech isn’t filmed like a YouTube video.

Vertical for social, the pace is faster and the sentences shorter. In 16:9 for YouTube you can breathe, and a shot that runs three seconds longer costs nothing. For a speech read from a stage, the text has to scroll noticeably slower, because you are addressing people who are present and you land the ends of sentences.

5. Do several takes, but not too many

Two or three takes are almost always enough. Beyond that, the tiredness is more audible than the improvement is visible.

If the third one is still bad, it isn’t a performance problem any more. It is the text that needs rewriting.

The tool I use

Running into points 2 and 3 is what eventually made me build Cue, a teleprompter that holds the script under the lens while the iPhone films, measures your real speaking rate with a thirty second test, then sets the scroll to match. It can also shorten a script that runs long or break it up for the voice, which is the work of step 2.

You don’t need it to apply any of the above. The method works with a printed page taped under a camera, which is what I did for years.

The summary

Text level with the lens. Written to be said rather than read. At a speed measured on you, not guessed. The rest fits into two takes.