Voice-controlled teleprompter

How voice control works, step by step

There is no calibration and no speed to rehearse. Voice control is three things happening at once, and understanding them tells you what to expect on a real take.

1. Live transcription. While the prompter is listening, your microphone audio is streamed to a speech-recognition engine and turned into text a few hundred milliseconds behind your voice. Only audio leaves your device — the script itself never does.
2. Word matching. Each recognized word is compared against your script around the position you are reading. A match advances the active word; a near-miss (a stumble, a mispronunciation, a filler word) is ignored rather than treated as a jump.
3. Scrolling. The text scrolls just enough to keep the active word centered, with damping so the movement is smooth rather than jerky. Spoken words fade so your eye naturally lands on what is next.

The practical result: you can slow down for emphasis, speed up through a list you know well, or stop entirely to think, and the script stays exactly where you are. Read the deeper explanation in what is a voice-controlled teleprompter.

Starts when you speak, pauses when you stop

This is what people mean by a voice-activated teleprompter, and it is built in. Press Present, start reading, and the text begins to move. Stop talking and it holds still — there is no timer pushing the next line at you.

• Silence is free. If you pause to breathe, take a question, or check a note, nothing scrolls. After 30 seconds of silence the microphone pauses itself so a long break does not use your speech time; one tap on Resume continues from the same word.
• Repeat a line and it finds you. Re-read a sentence for a cleaner take and the prompter recognizes the repeat and re-syncs to it instead of racing ahead.
• Skip a few words and it catches up. Drop a clause or jump a sentence ahead and the matcher looks around your current position (roughly fifteen words back and forward) to lock on again.

Hands-free by design: no remote, pedal, or operator

Because your voice is the control, your hands stay free for gestures, props, a guitar, or a coffee. That matters most when you record alone — there is nobody to run the prompter, and reaching for a remote is exactly the moment your eyes leave the lens.

• No clicker, foot pedal, Bluetooth remote, or second person
• No fixed scroll speed to set, test, and re-set for every script
• Runs on the laptop, phone, or tablet you already record with

If you present on stage, the same idea applies: the prompter sits on a confidence monitor or a phone near the lectern, and you never touch it. See the teleprompter for speeches and public speaking page for the stage setup.

Going off script — and coming back

Improvisation is the acid test for voice control. A prompter that treats every unscripted word as a possible match will wander; one that ignores you entirely is useless when you return. Sayscroll waits.

When you leave the script — a tangent, an example, a joke — the prompter notices that your words stop matching and simply holds position. It keeps listening. The moment you say the next scripted phrase, matching resumes and the text moves again. You can ad-lib for a sentence or for two minutes; the script is where you left it.

Practical tip: write your script the way you speak, in short lines, and mark the places you plan to riff. The how to use a teleprompter guide covers script formatting that makes re-entry effortless.

Accuracy, accents, and 60+ languages

Voice tracking is only useful if it works in the language and accent you actually speak. Sayscroll tracks speech in more than 60 languages — from English, Spanish, and German to Hindi, Arabic, and Korean — and you can mix languages inside one script. The full list, grouped by script with what to expect from each, is on the teleprompter languages page.

• Accents: recognition is trained on real-world speech, so American, British, Indian, or non-native English all track without a setup step. See teleprompter in English for accent details.
• Mixed-language scripts: a Spanish script with English product names, or a bilingual intro, keeps tracking across the switch.
• Word-level accuracy: in normal conditions tracking stays on the exact word about 99% of the time. A quiet room and a microphone within arm's reach are the two variables that matter most.

Numbers, abbreviations, and names are the usual edge cases for any speech engine. Writing "twenty twenty-six" instead of "2026" and spelling names the way they sound removes almost all of them.

Setup: microphone, distance, and eye line

A voice-controlled prompter needs two things from your setup: a microphone that hears you clearly and text placed close enough to the lens that reading looks like eye contact.

Text size, column width, margins, alignment, light or dark background, fullscreen (press F), and horizontal or vertical mirroring are all adjustable while you read. If you want to record in the same tab, turn on the built-in recorder and the take saves to your device — see teleprompter for YouTube for the filming workflow.

Try it in your browser

Paste a short script on the free teleprompter page, press Present, and read. No account, no download, no card. Not sure how long your script runs? The script timing calculator turns your word count into minutes before you hit record.