A teleprompter for TikTok, Reels and Shorts — say the hook exactly

On TikTok, Reels and Shorts, the first three seconds are the whole business. The hook is written — rewritten, tested, sharpened to one specific string of words — and it only works said exactly. “Approximately the hook” is a different hook, and the difference shows up as a swipe.

The format also runs fast. Short-form delivery sits well above conversational pace, and you cannot judge your own words per minute from inside your own head. Takes that felt energetic play back slow, and the feed does not wait for anyone.

Most of the feed watches on mute. A short with no captions is a mime act for a large share of its audience — the argument never arrives.

Retention compounds the hook problem. Short-form platforms grade a video on how much of it gets watched, so the middle cannot sag either: every beat is written to arrive at a timestamp, and a delivery that wanders adds seconds the script never budgeted. A sixty-word script has no room for an um. The caption burden is platform-shaped too — each feed lays its buttons and progress bar over a different slice of the frame, so text that sat safely on one platform is buried on the next, and captions can only respect each safe area if each platform gets its own export.

And it is a volume game. Daily or near-daily posting, with each platform preferring its own upload — which multiplies every manual step by every post, by every platform, by every week.

The standard method is memorise-and-repeat: say the hook until one take survives. Fifteen takes in, the words are finally right and the energy is gone — the accurate take is flat, and the take with life in it fluffed the hook.

The notes app propped above the camera is the other habit, and at selfie distance the eye-flick up to it is plainly visible. Viewers may not name what is off; they swipe anyway.

Batching is the standard advice, and it half-works: shoot seven in one session, schedule the week. But batching multiplies the takes problem by seven — a morning of memorise-and-repeat is an afternoon of a tired face — and the session collapses the moment the fourth script will not stick. The tools that schedule posts do nothing for the twenty minutes it takes to land one.

Captions get added in a second app, clip by clip, export by export — a daily tax on a daily poster. And one export re-used across all three platforms means someone’s safe areas crop your text.

Talk2Camera puts the exact words on the screen you are filming with. Voice follow tracks you on the device: hesitate and the script waits rather than running away; jump a line mid-flow and it catches up. The hook comes out as written, on the first take that has energy in it — not the fifteenth.

Cover mode fixes the eye problem a different way: the script covers the whole screen and the camera records behind the words. You are reading at the lens rather than watching yourself — which is also the cure for the self-conscious take.

After each take, the delivery metrics say what actually happened: pace in words per minute, filler-word count, and how much of the script you covered — measured on the device from the take itself. You learn your numbers. If your hooks land at 170 words per minute and the take says 128, you go again knowing exactly why.

Coverage matters as much as pace in a format this compressed. A take that covered 88% of the script is missing a beat, and in sixty words a missing beat is usually the reason the video exists. Seeing it as a number straight after the take catches the loss while the light and the energy are still in the room — not that evening in an edit, where the only fix is a reshoot with yesterday’s hair.

Captions can be burned into the picture, with style controls and per-line timing, so the muted feed still gets every word.

Then the export: presets for TikTok, Instagram and YouTube, the vertical feeds framed 9:16, each with its platform’s resolution cap, bitrate and duration guidance — one export per destination rather than one export stretched across three. On the free plan exports carry a small Talk2Camera mark; the paid plans take it off.

A creator posting daily writes three scripts on Sunday, sixty to eighty words each — hook, three beats, button. Monday morning, cover mode on: the script fills the screen, the camera rolls behind it.

Take one of the first script comes back at 131 words per minute, five fillers, 88% of the script covered. Too slow for a fast-cut feed, and the hook lost a word. Take two: 168 words per minute, one filler, all of it — done in four minutes, because the feedback was numbers rather than a feeling.

Captions burned in, then three exports: TikTok 9:16, Reels 9:16, Shorts. Posted by ten o’clock. The hook in each one is the hook that was written, word for word, because it was read — not remembered.

Multiply the morning by a week and the arithmetic is the argument: five posting days at three takes each instead of fifteen, nothing uploaded twice, nothing cropped by a feed the export was not made for — and the pace number drifting upward as 170 words a minute stops being a readout and becomes an instinct.