A voice activated teleprompter that follows the words you say.

What “follows your voice” means

Voice mode recognises your speech and scrolls to the words you actually said. Pause to breathe or let a car go past, and the script waits with you. Speed up and it keeps up. Say an earlier line again and it scrolls back to that line.

That last one shows the tracking is positional. The app matches what it heard against where you are in the text, so going backwards is as ordinary as going forwards and there is no drift to correct. There is no 3-2-1 countdown either: once the start delay is over, Voice and Sound wait for your first words.

Why a fixed speed gives you two jobs at once

A fixed-speed prompter asks you to do two jobs at once: deliver the line, and manage your pace against a machine that has no idea what you just said. On camera, people rarely notice the reading itself. They notice a pace that is slightly too fast or slightly too slow, and eyes flicking sideways to find the place. Following the words removes the second job.

Sound mode, and when to pick it

Sound mode moves while it hears you and stops when you stop. It does not match what it heard against the script. Two situations call for it. The first is a language your iPhone cannot recognise. Sound works in any language, because it is not listening for one. The second is a condition where recognition would struggle: a noisy room, a distant microphone, a script full of names and numbers. The trade is that Sound does not know you repeated a line, so it cannot scroll back on its own.

Constant, and the microphone hand-off

Constant is the conventional teleprompter: a fixed speed in words per minute, kept because sometimes it is the right tool. It is the only mode that does not need the microphone to scroll. Every mode has the same start delay: off, 5, 10 or 20 seconds before the script moves, shown inside the floating window as well as full screen.

Voice following keeps up while you record in Camera, Instagram and TikTok, the three apps that are tested. If some other app does take the microphone mid-take, the take waits for it. The text holds on the word you reached and a note says “Mic busy”, and reading carries on from that word when the microphone comes back. Constant is the mode to pick when you would rather nothing depended on the microphone to scroll. The floating window page has that hand-off in both directions; all three modes sit side by side on the home page.

Where the recognition runs

Recognition runs on your iPhone wherever your language supports it: Wispim asks iOS for the on-device recogniser whenever one exists. Where your language has no on-device model, iOS sends the audio to Apple to convert it to text. That is Apple’s pipeline, not ours, and the audio does not reach us either way.

You are told before a take, in your own language: the language picker marks languages that use Apple recognition. A full-screen notice can appear if on-device recognition repeatedly fails and the app falls back to Apple; the floating window shows no notice. You will not read “offline teleprompter” here, because for some languages that would be untrue. Your scripts, the audio and the transcripts are never sent to us; the privacy policy has the rest.

Three language counts, and they are different numbers

When it mishears you

It will, occasionally: a homophone, a name, a sentence lost to a passing truck. Recovery is what makes a following prompter usable, and none of it asks you to stop recording and start the take again:

• Tap a word. Reading carries on from the word you tapped, in every mode, and it never pauses.
• Drag to look around. Drag the script to read ahead or check a line you passed. Following pauses and a note at the top says “Browsing”; tap it, or anywhere off the text, to go back. In Voice you can also start reading where you landed, and Wispim finds you.
• Spoken commands. Pause, continue, go back a sentence, start over or jump to the end by saying so. They work in ten languages and the phrases are editable.
• Volume buttons. They move one sentence back or forward. They are marked Experimental in the app and do not work inside camera apps, which take the buttons.

Tapping and dragging are full-screen tools: the floating window does not accept touch at all, an iOS constraint and the reason the spoken commands exist. Choose your mode before a take that will live in that window, over Camera or over a Zoom or Meet call.