Teleprompter UX: keep scroll and record separate

Keep the reading position stable when text changes

Teleprompter.com combines font and alignment controls with independent horizontal and vertical mirror switches. Teleprompter for Video places a font-size slider beside Off, Horizontal and Vertical mirroring choices. These controls address the reading surface, but their relationship to the camera preview needs to remain explicit.

Anchor the current reading position to a stable passage or script element, not only a scroll offset. If larger text adds lines, preserving a pixel position can jump to another sentence. Keep the reader’s place through supported font, width and orientation changes, and make returning to the start an intentional action rather than a side effect of formatting.

Apple’s Pages presenter-mode documentation treats text appearance and automatic scrolling as configurable presentation features. In a camera app, clarify additionally whether a mirror setting changes the script, the preview or the recorded image. Those surfaces should not inherit one another’s transforms without an explicit product rule.

Give scrolling speed a meaningful unit and scope

Teleprompter shows a speed value beside a tortoise-to-hare control, elapsed and remaining counters, and a distinct START button. BIGVU displays Scroll Speed 160 above a record timer. The visible number alone does not establish words per minute or a shared speed scale between products.

Use an explicit unit where the implementation supports one, or a clearly explained relative scale. Font size, line spacing and script width can change how quickly readable words pass the guide even at the same pixel speed. Do not imply that one saved number guarantees the same delivery pace after every formatting change.

Let people rehearse without creating an unwanted recording, and distinguish speed adjustment from seeking through the script. If the product changes speed while recording, keep the new effective state visible without covering the next line. A reset should explain whether it resets the rate, the reading position, the timer or more than one of those values.

Separate choosing a countdown from counting down

VILO shows a Countdown duration panel over its camera and script workspace. Captions shows a large countdown numeral over a capture view that also contains a stop-shaped control and take-related information. A settings value, an active countdown and active capture are different states even when they share the same screen.

Explain when capture actually starts and what a countdown cancellation preserves. A person should be able to stop preparation without accidentally discarding a previous take. Do not use one ambiguous timer for time until recording, time already recorded and estimated reading duration; name or position those quantities so their roles remain understandable.

Check what happens when the app is interrupted before zero, when the camera cannot start or when the microphone is unavailable. Avoid continuing the script as if capture succeeded. Preserve the prepared position and state the failed operation. A visible countdown reaching zero is not evidence that the recording device actually created a usable file.

Make voice following different from fixed-speed scrolling

Vozo offers a Voice-Synced Mode switch with copy promising automatic movement in response to speaking pace. Dubs presents a Script Speed control labeled Fast. One advertises a following behavior; the other exposes a rate setting. A recorded settings screen does not establish how reliably either mode follows an actual performance.

Keep the active mode obvious and explain any additional permission or processing requirement at the relevant moment. Microphone use for recording and microphone use for speech following serve different roles, even if they share device access. Do not suggest the reader can rely on following before the required capability is available.

Provide a recoverable response to silence, a repeated phrase or manual seeking. If manual scrolling suspends following, show how it resumes. Keep the script and current passage available if recognition fails rather than restarting everything. Test the supported languages and speaking conditions without presenting one successful rehearsal as proof that all delivery styles will work.

Keep the accepted take separate from the prompting surface

VivaVideo offers Rolling Speed and text settings above a review area with Retake, playback and Use Video. AI Cut exposes Suspended Words with font, speed and appearance controls before an Open floating window action. The first surface can accept a recorded artifact; the second prepares another presentation context.

Name the point at which a take becomes the chosen video. Stopping a recording should not automatically mean discarding it, and changing the script should not silently overwrite an accepted take. Keep the relationship between the script revision and recorded take clear enough that someone can compare versions without guessing which words were used.

Treat a floating prompting surface as a capability with lifecycle boundaries, not an unconditional always-on promise. Explain supported contexts and provide a return path if the destination app, window or permission is unavailable. Closing that surface should not secretly delete the script or invalidate the previous recording; define and test those persistence rules separately.

Model script movement and capture independently

Represent script identity and revision, reading anchor, scrolling mode, capture lifecycle and take identity as separate state. A paused reader can coexist with active capture. A returned camera failure must update capture status without pretending that a script animation also failed or succeeded. Avoid deriving the entire workspace from a single playing boolean.

Keep the displayed recording state tied to the capture system’s confirmed lifecycle, not only the button press or countdown completion. Retain a recoverable take until its replacement is accepted. When script text changes, preserve meaningful anchors where possible and identify when the current take belongs to an earlier revision instead of silently relabeling it.

Expose named controls, readable focus order and non-drag alternatives for supported precise adjustments. Give screen-reader users access to script content and recording status without announcing every scroll tick. Consider motion sensitivity and rehearsal needs, but do not promise a static preview validates camera permissions, audio capture or the behavior of another app’s floating-window context.

Rehearse interruptions, not only a perfect take

Begin rehearsal, change the font, mirror the text and pause scrolling mid-paragraph. Ask the tester to find the same passage and explain whether a video exists yet. Then start a countdown, cancel it and inspect the prior take. This separates reading-position failures from misunderstandings about what the camera has saved.

Record a short supported script, manually scroll backward, interrupt capture and return. Test a failed microphone request, unavailable camera, long script and large text. For voice following, include silence and repeated wording; for floating text, leave and reopen the supported context. Inspect the saved video and audio rather than trusting the preview timer.

Measure lost-place incidents, unusable takes, mistaken discard actions and capture failures separately from sessions started. A longer rehearsal can mean preparation or confusion, not better performance. Review whether people can predict the next action, recover the current passage and retain a usable take without reconstructing their entire setup.

Review the reader, recorder and saved take

Use these checks on a working build, not only on the polished reference state.

• ✓The current script passage survives supported layout changes.
• ✓Mirror controls identify the surface they transform.
• ✓Scrolling rate and reading mode are understandable.
• ✓Countdown, script movement and actual capture are distinct states.
• ✓Manual seeking and voice-following recovery are explicit.
• ✓Stopping capture is separate from discarding or accepting a take.
• ✓Interruptions preserve the script and any recoverable recording.

Connect preparation to the rest of video creation

The AI video creation examples cover the wider creation journey. A teleprompter needs its own reading and capture model before a take enters that editing workflow.

Use the mobile color and typography examples to evaluate readable script treatment, then test it at the real viewing distance and supported camera setup.