Voice-Controlled Teleprompter: How VoiceScroll Works

Regular teleprompters scroll at whatever speed you set. You press play, the text moves, and you try to keep up. If you've ever used one, you know the feeling: you're either racing to catch the text or waiting awkwardly for it to catch up to you.

Voice-controlled teleprompters work differently. They listen to what you're saying and move the text to match. Here's how that actually works in VoiceScroll.

The problem with fixed speed

People don't talk at a steady rate. You speed up when you're comfortable with the material, slow down when you hit something important, and pause between ideas. Within a single sentence, your pace can vary by 30% or more.

A fixed-speed teleprompter ignores all of this. The result is that you end up performing for the machine instead of speaking naturally. Your delivery flattens, you rush through pauses, and you re-record more often.

How voice matching works

When you load a script into VoiceScroll, the app does a few things:

1. It breaks the script into individual words and builds a matching index.
2. When you tap Start, the microphone turns on and Apple's speech recognition starts processing your audio.
3. As you speak, the app converts your speech to text in real time and matches the recognized words against your script.
4. The current word highlights in orange. Words you've already said fade to gray. The view scrolls to keep your position visible.

The whole loop runs on-device. Your audio stays on your phone.

On-device processing

Everything happens locally on your device using Apple's speech framework. No audio goes to a server. No internet connection needed after initial setup. No API call limits.

This also means there's no network delay. The matching feels instant because it is.

Language support

VoiceScroll works in 9 languages: English, Korean, Japanese, Chinese (Simplified and Traditional), German, French, Portuguese, and Spanish. Each language uses Apple's native speech recognition model for that language, not an English model with translation bolted on.

This makes a real difference for languages like Japanese and Chinese, where word boundaries, pronunciation patterns, and character handling are completely different from English.

Where it works best

• Live presentations where you might get interrupted or need to pause
• Content in multiple languages
• Long recordings (sermons, lectures, audiobook narration) where your pace naturally drifts
• People who are new to teleprompters and haven't learned to match a fixed scroll speed