Voice-activated teleprompter: how it works
A traditional teleprompter scrolls at a fixed speed, and you adapt to it. A voice-activated or voice-following teleprompter does the opposite: it listens to you and adapts to your speech. When you pause to think or stumble on a word, the script waits. This guide explains how that works, what it needs and where its limits are, using Telepromptia's FlowSync as the example.
The problem with fixed-speed scrolling
With a fixed speed you have to match the machine. If you speed up, you run off the end of the visible text. If you pause, the words scroll away and you have to find your place. To avoid both, people read in a steady rhythm that sounds like reading. Voice following removes the constraint by letting the position of the script depend on what you actually say.
Note that this is different from voice commands such as saying "start" or "stop". Voice following listens to the words of your script, not to commands.
How voice following works
The idea has three steps.
- Listen. Speech recognition turns what you say into a stream of words.
- Match. The app compares those words with your script and works out which sentence you are on.
- Scroll. The script moves so that the current sentence sits in the reading zone, and it stops when you stop.
Real speech does not match a script exactly. People add filler words, say a phrase twice, change a word or skip a sentence. A useful engine has to ignore capitalisation and filler words, accept small slips and recover when you jump ahead. FlowSync's matching logic is built for these cases, and it has automated tests with scripted transcripts covering fillers, slips, a skipped sentence, a repeated phrase and a long jump.
When it hears something it cannot match, such as a cough, a side remark or unrelated speech, it holds its position instead of guessing. That is the safer failure: the script stays where you left it, and you can continue from there.
Speech recognition on the device
Telepromptia runs voice following with on-device speech recognition only. Your voice is not sent to a server for this feature, and the app never switches to server-based recognition as a fallback. That also means it works without a connection. If you care about privacy, this is the part to check in any app: ask whether the recognition runs locally.
What it looks like in practice
Say your script opens with "Today I will show you three ways to save time." You begin with "so, um, today I will show you three ways to save time". The filler words are ignored and the highlight moves on to the next sentence. You stop for five seconds to check a note, and the script waits. You come back and skip the second sentence because you already said it earlier. Once the engine is confident that you are on the third sentence, it moves there instead of dragging you back.
What it needs
On-device recognition is not available for every language on every device. Telepromptia switches FlowSync on only when Apple reports that on-device recognition exists for the language you chose. If it does not, the feature is not offered for that language and you can use timed or manual scrolling instead. Voice following is a Pro feature in Telepromptia.
Tips for better results
- Use a quiet room. Background speech and loud noise can confuse the matching.
- Speak at a normal pace. You do not need to speak slowly or clearly like a robot. Talk as you usually do.
- Write words the way you say them. Numbers, abbreviations and unusual names are easier to match when the script spells them as you pronounce them.
- Keep the phone at a normal distance. The microphone should hear you clearly without clipping.
- Rehearse once. Voice following helps with timing, but you still need to know your material.
Results depend on your voice, the room and the language, so try it on a short script before you rely on it for a long one.
When fixed speed is better
Voice following is not always the right choice.
- If the room is noisy or other people are talking, fixed or manual scrolling is more predictable.
- If your language has no on-device support, you need the fallback anyway.
- If you have rehearsed the script heavily and speak at a very even pace, a timed scroll works fine and needs no recognition at all.
- If you want a script to fit an exact length, a target length setting gives you control. Telepromptia can set the scroll speed so that the script lasts 30, 60 or 90 seconds.
How it pairs with other features
Voice following tells you where you are, and sentence highlight shows it. If you make a mistake mid-take, MagicRetake lets you re-record just that sentence and keeps your other takes. Together they make a long script manageable, as we describe in the guide to using a teleprompter on iPhone and the guide on fixing a mistake without re-recording everything.
Where Telepromptia fits
FlowSync is part of Telepromptia Pro. The free plan includes the scrolling teleprompter with speed, size and colour settings, EyeLock layouts and clean export with no watermark. There is no account, and scripts and videos stay on your device. For details see features and pricing, or read how we handle data in the privacy policy.