STOP READING AT A MACHINE'S PACE Anyone who has recorded to a teleprompter knows the feeling. The text is moving, you're not quite with it, and every take becomes a negotiation between how you want to say the line and how fast the words are sliding past. Speed it up and you gabble. Slow it down and you drift. The performance ends up serving the equipment. Onwords removes that negotiation. It listens through your microphone, works out where you are in your script, and scrolls the text to keep your current line centred on screen. You speak at whatever pace suits the material, and the script keeps up with you. No clicker. No operator sitting off-camera. No guessing. FORGIVING BY DESIGN Real delivery is never a clean read-through. You'll stumble, repeat a phrase, drop in an unscripted aside, or jump a paragraph because the previous take already covered it. Onwords is built for that. It needs only a few recognisable words to hold its place, so a fluffed line or a filler word won't throw it. When it isn't confident about what it heard, it holds position rather than guessing and leaping somewhere wrong. And when you do move somewhere unexpected, it searches the whole script and picks you back up within a sentence or two. YOUR SCRIPT NEVER LEAVES YOUR COMPUTER Speech recognition happens on your own machine, using the engine built into Windows. There is no cloud service, no API key, no account to create, and no internet connection required at any point. No audio is recorded or stored. Nothing is uploaded. That matters when you're working on something under embargo, under NDA, or simply not finished yet. TWO PRESENTERS, ONE SCRIPT Co-hosted shows, interviews and double-act pieces all break a normal teleprompter the same way: one undifferentiated block of text, with no warning that your line is next. Onwords lets you assign lines to either presenter and tints them on screen - pale yellow and pale blue - so a handover is something you see approaching rather than something you discover. Tracking carries straight through the changeover, because it follows whoever is talking. Both presenters share one microphone. Assignments are stored in the script itself, so they survive editing, they save with your script, and a script that already marks up its speakers works as soon as you paste it in. BUILT FOR HOW PEOPLE ACTUALLY FILM Mirror the text horizontally for beam-splitter teleprompter hardware. Scale it up for distance from the screen, or down to fit more on. Run it full screen on a second monitor so the prompter display stays clean while you keep the controls on your laptop. A live transcript shows you exactly what the app is hearing, so nothing about the tracking is mysterious. The arrow keys nudge your position by hand at any moment, and the mouse wheel scrolls freely whenever you want to look ahead. And if you'd rather not use your voice at all, a classic constant-speed scroll with an adjustable rate is still there. A REAL WINDOWS APP Onwords is built with WinUI, the current native Windows interface framework. It starts quickly, sits at almost no CPU while you read, and doesn't drag a browser engine along with it. Native builds for both Intel and ARM processors. Your script is saved automatically and restored next time you open the app, so there's no re-pasting between takes or between days. WHO IT'S FOR Video creators recording to camera. Corporate and marketing teams producing internal and customer-facing video. Speakers and presenters delivering keynotes. Educators and course creators recording lessons. Livestreamers and webinar hosts who can't stop to find their place. Podcasters working from a scripted structure. And anyone recording alone, with nobody available to run the prompter for them. A NOTE ON ACCURACY Onwords uses the speech recogniser built into Windows, which will mishear unusual names and technical jargon. Onwords is designed to track well in spite of that, rather than to transcribe you accurately. It follows your position in a script you have already written — it is not a dictation tool.