Open the site and start typing — no signup, no install, no cloud. Your script scrolls at your pace, waits when you breathe, and forgives the fifth take.
Hey everyone — welcome back to the channel.
Today we’re building something I’ve wanted for years.
A teleprompter that actually listens while you speak,
and never scrolls off without you.
Tap the mic and read. The script follows your pace, waits when you pause, and glides — never jumps — back up to speed. Ad-lib freely; when unsure, it calmly holds position instead of guessing.
Open the same link on your phone — no pairing, no app. Play, sentence jumps, speed, font size, and a live progress bar you can scrub. Changes land on every screen within a keystroke.
Speech recognition runs 100% on-device — a ~28 MB model, cached after one download. Zero third-party requests, unguessable script URLs, works offline once loaded.
Your script and its private link exist the moment you type. That’s the whole onboarding — the URL itself is the key, with ~83 bits of randomness standing guard.
Share it and anyone can edit or drive the prompter — collaborators, camera ops, your phone. No accounts needed.
Sign up when you want a script list with previews and word counts. A script written anonymously is claimed with one click.
Edits persist automatically and sync to every open device within a keystroke. Pause mid-take, fix a line, keep rolling.
Live WebSockets keep text, play state, position and settings in lockstep — with automatic reconnection if the network blips.
One tap jumps to the start of the sentence you flubbed, parked right at the reading line and ready to re-read. Press back again for the previous one — track-skip, but for talking.
Every sentence is its own stop, even three-to-a-line. And a standard slide clicker works out of the box: PageUp / PageDown are wired as your retake remote.