If your teleprompter keeps stopping mid-take, the cause is almost always the speech engine timing out on long pauses. Most teleprompter apps run their voice tracking through a cloud speech service or your browser’s speech engine, and both time out after about a minute of silence — or the moment your Wi-Fi drops. So you pause to sip water, reset a take, glance at your notes, and the app shuts off the mic. When you start speaking again, nothing scrolls. This post covers exactly why this happens, the two fixes that actually work, and what to look for in a teleprompter built to handle real shoot-day pauses.

Why teleprompters stop in the first place

Voice-tracked teleprompters work by streaming your microphone audio to a speech recognition engine, which converts it to text in real time and matches that text against your script. Your scroll position updates based on where the match lands. The whole pipeline depends on the speech engine being active and listening.

The problem is that most of the speech engines teleprompter apps use weren’t designed for the long silences that happen on a real shoot. They’re optimised for dictation — short, continuous bursts of speech. When you go quiet for 30, 60, or 90 seconds between takes, the engine interprets that as “the user has stopped dictating” and closes the session to save battery and processing cost. When you start speaking again, the session has to restart, and most apps fail to notice that’s happened.

Which speech engines have this problem

Web Speech API. The worst offender. This is the engine powering most browser-based teleprompters, and it times out after roughly 60 seconds of silence on Chrome. It also requires an active internet connection — if your Wi-Fi has a wobble mid-take, the session dies.

Cloud speech recognition. Services like Google Cloud Speech-to-Text and AWS Transcribe work similarly — they stream your audio to a server, get back text, and close the connection on inactivity. Adds latency on top of the silence timeout, and again, no internet means no tracking at all.

On-device speech recognition. Engines that run locally on your phone or tablet — like Apple’s SpeechAnalyzer on iOS and the equivalent on Android — work offline, have much lower latency, and (when implemented properly) handle pauses without dropping the session. The catch is that “implemented properly” is the part most apps get wrong.

Fix 1 — switch to a teleprompter with on-device voice tracking

This is the real fix. An app that runs speech recognition natively on your device, with proper handling of silences and session resumption, won’t stop on you mid-take.

The quick test for whether your current app is on-device or cloud-based: put your tablet or phone in airplane mode and try voice scrolling. If voice tracking still works, you’re on-device — your problem is somewhere else (worth checking microphone permissions, language settings, or how closely you’re sticking to the script). If voice tracking stops the moment you go offline, you’re on a cloud-based app, and the speech engine is the bottleneck. Switching apps will fix it.

Steady Cue runs voice tracking natively on your device. Long pauses, take resets, water breaks — the session stays open, the script stays exactly where you left it, and tracking picks up cleanly when you resume.

Fix 2 — add a Bluetooth remote as a backup

If you can’t switch apps right now, the workaround is a physical fallback. A Bluetooth remote (around £20 for something like a Neewer presenter remote, or built-in support if your app already handles them) lets you nudge past a stopped scroll without waiting for the speech engine to wake up.

This isn’t a fix for the underlying issue — every time the engine drops, you’re still manually overriding it — but it’ll save the shoot day. Most voice-tracked teleprompters that support voice tracking also support Bluetooth HID remotes; check your app’s settings. If yours doesn’t, that’s another signal it might be worth switching.

A note on external microphones

One extra cause worth mentioning: Bluetooth lavalier microphones (Rode Wireless GO, DJI Mic, and similar wireless lavs) can break voice tracking on some apps because of audio latency. The mic adds 50–200 milliseconds of delay, which can desync the matching engine and cause the app to stop because it can’t find your position in the script. Wired lavaliers usually work better. If your voice tracking started failing after you added a wireless mic, try switching to a wired one as a diagnostic.

The bottom line

Almost every “teleprompter keeps stopping” problem comes down to the speech engine timing out on pauses. Switch to an app with proper on-device voice tracking, or layer a Bluetooth remote in as a fallback. Both solve the problem; the first solves it permanently.