Teleprompter for YouTube: setup and takes

What a prompter does in a YouTube workflow, and what it does not

It scrolls your script past a fixed line at roughly the speed you talk, so your eyes stay near the lens instead of dropping to a page.

It does not record. Your camera records, or your capture software does, and nothing about the video passes through this page. There is no export, no watermark question and no upload, because there is nothing here to upload to. Your script stays in your own browser.

Everything below assumes that split: the prompter handles the words, and the thing you already record with handles the video.

Decide the length before you write the script

Words per minute is the practical way to guess how long a video will run, and it is the same number the roll uses. At 130 words per minute, the setting the stage opens with, an eight minute video is roughly 1,040 words. A 1,500 word script at 140 words per minute takes about ten and a half minutes to read start to finish.

The stage shows that duration next to the transport controls, worked out from the speed you have actually set rather than a generic average. Paste a draft in and you know immediately whether it is a six minute video or an eleven minute one. If you would rather work out the length before you get near the prompter, the script timer runs the same arithmetic on a page of its own.

The number bends in both directions once you start shooting. Pauses, retakes and breathing make the raw read longer than the estimate, and cutting those pauses in the edit makes the finished video shorter than the read. Treat the estimate as the length of the words, not the length of the video, and leave yourself a margin either side of the length you are aiming at.

Where the prompter sits relative to the lens

A read shows in the eyes when the script is a long way from the lens, and mostly stops showing once it is close. A phone or tablet directly beside the camera at eye height is enough for most channels. A beam splitter puts the text on the lens axis itself, which is why studios use them and why mirror mode exists.

The mechanics for each screen are on their own pages: iPhone for a phone in landscape beside the camera, iPad for a tablet on a stand or in a splitter rig, Mac and Windows for reading off the machine you are recording with. Setting up the glass is on the mirror mode page.

Record it in sections

A ten minute video read in one unbroken take is the hardest version of this job, and the finished video almost never needs it. It gets cut anyway.

The dock has a script switcher and a button for starting a new script, so a video can be five short scripts rather than one long scroll: the hook, the setup, two segments, the outro. Each one gets its own duration readout. Each one is short enough to fluff and redo without hunting for your place, and recue puts you back at the top of the current one in a single click.

Between takes, the cut hides the seam, which is how most of the confident pieces to camera you have watched this week were actually made.

The first fifteen seconds are where a read shows most

The opening is where people decide whether to stay, and it is also where reading is most visible: nobody has settled into the video yet, so they are looking straight at your face.

The simplest fix is to not read the opening at all. Say the first two sentences to camera from memory, then press Play and let the script take over underneath you. Two sentences is a small thing to learn and it buys the whole opening.

If you would rather keep the hook in the script, rehearse it two or three times at the speed you have set, then run it with the three second countdown so you are looking at the lens before the first word moves.

Set the speed to the way you actually talk

Most people talk somewhere between 120 and 160 words per minute on camera. Start in the middle, read a paragraph, and adjust while reading rather than in the abstract. The speed is in words per minute rather than pixels per second, so the setting means the same thing when you change text size or move to a different screen.

If you cut tightly in the edit and want room to breathe, run slightly slower than your natural rate and leave the pauses in; they cost nothing because they are coming out later. If you publish close to what you shot, match your actual talking rate and stay there.

Write the script so it survives being read aloud

The stage shows plain text. There is no bold, no highlighting and no import step, so a script pasted out of a document arrives as words and nothing else.

• Write numbers the way you say them. "Twenty twenty six" and "about a third" are readable at speed; "2026" and "34%" make you do arithmetic out loud.
• Spell awkward names the way they sound, in brackets after the first one if you need the real spelling in the description.
• Keep paragraphs to two or three sentences and leave a blank line where you breathe. Those blank lines are where you can look up.
• Use a line break for emphasis, because that is the only emphasis the stage has. A short line on its own lands harder than a bold word would.
• Read the draft aloud once and cut whatever you trip on. Anything you stumble over twice was never going to survive a take.

Shorts and vertical video

Shorts are short enough that the arithmetic is unforgiving. At 140 words per minute, a 45 second Short is about 105 words, and a 60 second one is about 140. That is one idea, said once, with no preamble.

Shooting vertical usually means a phone is doing the filming, so the script has to live on something else: a second phone, a tablet, or a laptop propped behind the camera. You cannot film on a phone's front camera and read from that same screen, because the camera app is in front of everything. Keep the text large and the script tiny. There is more on the format in the Shorts post.

Tutorials, demos and anything that shares the screen

If your capture includes the display the prompter is on, the prompter is in the video. That never comes up in a talking-head shot, because you are capturing a camera, and it comes up in the first minute of a screen-recorded tutorial. Capture a single window, use a second monitor, or move the script onto a phone beside the lens. The Mac and Windows pages go through both arrangements.

What ends up in the video that you did not plan

• A bright screen shows up in glasses. A dark stage with light text mostly solves it, and both colours are adjustable from the dock.
• Long lines make your eyes track sideways, and the flick at the end of each one is visible. Raise the text size until lines break sooner.
• A long sentence drains the energy out of a delivery, because you are managing it rather than saying it. Shorter sentences read livelier without any change in performance.
• A locked head and a fixed gaze read as reading even when the words are perfect. Move your hands. Look away deliberately at the end of a thought, the way you do in conversation.

Questions

Does it record my video? No. It scrolls the script; your camera or capture software records.

Does my script get uploaded to YouTube, or to you? Neither. It stays in your browser on your own machine.

Do I need an account? No. There is no sign up and nothing to upgrade to.

How long should the script be for a ten minute video? Around 1,300 words at 130 words per minute, before you cut anything. Paste your draft into the prompter and it will tell you exactly, at whatever speed you actually read at.