← All notes

How to read a teleprompter naturally

The tell is never the reading itself; it is the pace and the eyes. We film presenters every week, most of them not presenters at all, and the same few fixes work every time.

Nobody minds that you are reading. Television newsreaders read from one for a living, and it bothers no one. What gives people away is what reading does to them: a pace locked to the scroll, eyes sweeping line to line, and a voice that flattens because the machine is in charge. Fix those three and the reading disappears. Here is how we set presenters up on our own shoots, using Miracue, the free prompter we built for those shoots.

Why you sound like you are reading

A traditional teleprompter scrolls at a fixed speed, and something subtle happens the moment it starts: you begin serving the scroll. You rush a line to catch up, stretch a word while the text arrives, and stop breathing where you would naturally pause, because the machine does not pause. The delivery flattens into the sound everyone recognises, and it has nothing to do with your ability on camera. The person is following instead of speaking.

Everything below either takes the machine out of charge, or shrinks what the viewer can see of the mechanics.

Write it the way you talk

Most "teleprompter voice" is actually writing voice. A script drafted like a document reads like a document, no matter who performs it. Before any camera is involved: read the script aloud once, and change every phrase you would not say across a table. Short sentences. Contractions left in. If a line makes you take a breath halfway, break it into two.

Then leave the paragraph breaks alone, because white space is where your breathing lives. A wall of text pushes a reader on; a script with air in it gives you permission to pause, and the pauses are what sound human.

Set the text so your eyes barely move

The second tell is eye travel: the visible sweep from line to line, and the flick back to the start of the next one. You cannot act it away, but you can shrink it below what a viewer notices.

Make the text bigger than feels sensible, from where you actually sit or stand, not with your nose at the screen. Bigger text means fewer words per line, and fewer words per line means smaller eye movements. Keep the column narrow, for the same reason; prompter text is set tall and narrow on purpose. And put the words as close to the lens as they will go: directly under the webcam, or reflected in front of the lens on a proper rig. The reading line in Miracue sits near the top of the screen so your eyes rest at their highest, closest point to a camera above it.

Miracue's prompter showing a founder script in large text in a narrow column, with the reading line marked at the left edge.
Reading size in practice: large text, a narrow measure, and the eyeline marker where your gaze should rest. This is the app as it ships.

Distance helps more than people expect. The further you are from the camera, the smaller the angle your eye movement makes, which is one reason a filmed piece to camera is usually shot from further back with a longer lens: at that distance, reading simply stops being visible.

Let the prompter follow you

This is the one that changes the delivery, because it removes the cause instead of hiding the symptom. Miracue's voice follow listens on your device and moves the script to match what you actually say. Speak faster and it keeps up; slow down to land a line and it slows with you; stop entirely, to breathe or to think, and it waits. Go off the script to add something and it holds still until you come back; re-read an earlier sentence and it finds you there.

The pace problem disappears, because there is no pace to serve. You are speaking, and the script is the one doing the following. On a fixed-speed prompter the equivalent discipline is to set the speed slower than feels comfortable and let it push you less; but if your device can run voice follow, run voice follow.

A practice routine that works

From our own shoot days, with presenters who had never stood in front of a camera before that morning:

  1. Read it aloud once for meaning, away from the camera, and fix what your mouth trips on. This is where the script stops being a document.
  2. Do one take you intend to throw away. Nobody's first take is their best and knowing it does not count takes the pressure off. Treat it as the rehearsal the camera happened to see.
  3. Then perform to the lens, not the words. The words are close enough to read with your eyes parked at the camera; talk to the person you imagine behind it. On our shoots the third take is usually the keeper: warm by then, not yet tired.
  4. Film in sections. Break the script into chapters and take them one at a time, jumping back for retakes. A ninety-second section read well beats a seven-minute read survived.

One more piece of direction from set: read half a line ahead of your mouth. It sounds impossible and is not; it is how everyone reads aloud well, and it is what lets the thought arrive before the words do. Pair it with deliberate emphasis, one word in each sentence that carries it, and the read stops being flat without any acting at all.

How fast should it go?

At the speed you would talk without it. A comfortable on-camera read sits around 140 words a minute, which is what Miracue's reading-time estimate assumes when it tells you a script's length, and it is a useful sanity check: if your two-minute idea is 500 words, the script is the problem, not your delivery. But the number is a reference, not a target. The right speed is yours, which is the whole argument for a prompter that follows your voice.

A cinema camera on a tripod with an iPad mounted directly beneath the lens, rigged for a prompted piece to camera.
On our own shoots: the script rigged directly under the lens, so the presenter's eyes never leave the camera's neighbourhood

Reading on a call instead of a camera?

The same craft applies on Zoom, with one change: the camera is at the top of your screen, so the words dock directly beneath it, and Miracue's call mode can pin the script in a little window that stays on top of the call, in Chrome or Edge on a computer; every free account includes three pinned calls, and Pro keeps the pin. The full setup, including what reads as eye contact on a video call, is in a teleprompter for Zoom calls.

Trying it

Practise with the free teleprompter: paste two minutes of something you actually have to say, make the text bigger than feels sensible, and do three takes. Throw the first away on purpose. By the third take the reading usually stops showing; that is not a promise about talent, just what we watch happen on set most weeks.

Read next

How to record a video while reading a script covers the full camera setup, and what is a teleprompter? covers the eyeline and the glass in more depth.

Written 4 September 2026 by Lumira Studio, a video production company in Hertfordshire, United Kingdom. The advice above is how we actually direct presenters, written down.