Everything here is stored on this device only — no account, no sync, nothing to sign into.
Script skin — the color scheme you read by. Six swatches, all measured at 11.5:1 contrast or better: Classic white-on-black, Studio Amber (warm, low blue light), High-Vis Yellow, Terminal Green, Paper (light, for bright rooms), and Slate (soft dark). The default is Classic; you can also switch skins mid-read with the 1–6 keys.
Script font — six Windows typefaces in legibility-first order: Segoe UI (default), Verdana (the widest, reads best at a distance), Trebuchet MS, Tahoma, Arial, and Georgia if you prefer a serif.
Scroll glide (100–320 ms) — how quickly the text glides when it moves to the next line. Lower is snappier, higher is more cinematic. Motion is always a smooth glide, never a teleport.
Guide band position (10–50% from the top) — where the highlighted band sits. Put it as close to your camera lens as you can: your eyes travel less, and less eye movement shows on camera.
The band marks the line you are speaking. The page scrolls up a few words before you reach the end of the line, so the next line is already waiting in the band — your eyes stay parked in one place instead of chasing the scroll.
Tracking model — the on-device speech-recognition model. base.en (default) is the accurate one. tiny.en runs about 3× faster with somewhat lower accuracy — good for fast readers or thin hardware. The choice applies the next time you start prompting. If base.en falls behind on your machine mid-read, the app switches to tiny.en automatically (one-way, and the status chip says so) rather than letting the scroll stall.
Recognition language — English (US & UK), Español, Français, Deutsch, Italiano, Português (BR).
Voice commands — the on/off switch for hands-free control ("teleprompter, pause" and friends). The full command set is on the Prompter & voice page. Off means pure voice-follow tracking — no command analysis at all.
Learn my wake word — if the app mishears "teleprompter", press this and say it once: the app records how the recognizer hears you, for each installed model, and matches that from then on. A green check shows something was learned (hover for details); Clear forgets it.
Input device — defaults to the Windows system-default microphone, so it follows whatever you select in Windows sound settings. Pick a specific device from the dropdown when you present on a different mic than usual.
Mic test — press Start mic test and read a sentence at your presenting volume. The meter shows live level with a peak marker, and three lights judge the take: signal detected, level healthy, not clipping. Thirty seconds here is the cheapest insurance against a silent recording.
Your activation email and (masked) key, seats in use, and how many days of offline grace remain — the app keeps working offline for a generous window between validations. Deactivate this device frees a seat so you can move the license to another machine; reactivation with the same key works any time.
The app checks for a new version once per launch and asks before downloading anything — the check only reads a version number from this website and sends no data. Check for updates runs the same check on demand. "Skip this version" is remembered per version. When an update installs, the app relaunches itself as soon as the installer finishes — no guessing when it's done.
Scripts, settings, and your voice are processed and stored on this device only. No accounts, no cloud, no telemetry. The only unprompted network calls in the entire app are license validation and the read-only update check above; a diagnostic log leaves the device only when you explicitly send it to support.
Contact support opens a short form inside the app — what you were doing, your reply email (prefilled from your license) — and attaches your latest diagnostic log if you choose. It goes straight from the app; no mail client needed. To record a fresh log of a tracking problem, press Ctrl+D on the prompter screen first, reproduce the issue, then send the report.
Back to: Using the app →