Speechius

I run a lot of live demo calls, and I never go through my talking points in the same order twice — whatever the other person brings up decides where I go next. Tap-to-reanchor sounds like it handles the occasional detour, but does it hold up if someone jumps between non-adjacent sections of the script repeatedly in one call, or is it really built for staying close to one linear path with the odd tangent? Either way, "the machine follows you" is the right framing — most teleprompters assume the human is the unreliable part.

The re-anchor tap when you drift off-script is the right escape hatch. Quick day-one question: what's the fastest way to get existing scripts in — does it take .docx or Google Docs imports, or is it paste-only? And can you have more than one script loaded at the same time, like if a presentation jumps between two topics and you want to switch mid-session without reopening the app?

The voice-following part is the clever bit. What happens when I pause to think or take an audience question mid-sentence? Does it hold position, or scroll ahead hunting for a match?

the part I actually care about is the off-script tracking, every teleprompter app handles reading verbatim fine but the second you ad-lib a joke or answer a question mid-webinar they lose your place. how does it re-sync once you go back to the script, does it match on nearby phrases or does it just resume from where you left off scrolling?

the deviation-matching answer above is thorough, so a different case - what happens when you just go quiet, not off-script but stopping to think mid-sentence or fumbling a word before continuing. does the cursor just hold at the last confirmed word until you pick back up, or is there any risk of it treating a long pause as "done with this line" and nudging ahead early?

Good one, I use to miss words while reading. This would really help for my my planned live sessions.

The part I'd actually pressure test is what happens when you deviate from the script rather than just paraphrase it closely, since that's the realistic failure mode, not clean word for word delivery. If you go on a tangent for thirty seconds and then land back on a sentence that's near the top of the script, does it correctly jump forward to match, or does it get confused and either stall on an old position or jump ahead to the wrong matching phrase somewhere else in the document. Fuzzy matching speech to a script is genuinely hard exactly at the moments when someone stops reading verbatim and starts talking naturally, which is presumably the whole reason you built this.

Staying hidden from screen recording and screen share is a neat trick, but that generally depends on how the OS classifies overlay windows, and that behavior isn't guaranteed to be stable across OS updates on either platform. Is that hidden from capture behavior something you're relying on a documented API for on both macOS and Windows, or is it more of a workaround that could quietly break the next time either OS changes how it handles window capture exclusion.

the "machine follows u instead of u following the machine" line is honestly the whole product. feels obvious once u hear it.

This seems to be pretty useful (I saved it into my "Interesting" folder). Might use it in the future :)

Hi everyone 👋

I'm Jay, a solo dev. Since 2020 I've presented online every day — YouTube videos and teaching programming over Zoom. And two forces have been fighting inside me the whole time.

The German force 😁 says: "Write a script. Stick to it. Nothing can go wrong."
The artist says: "Are you a robot? Where's your personal touch?"

Reading from a script gives me control, but it sounds read. Improvising sounds natural — until I lose my thread and the video drags on twice as long. I wanted the middle ground: speak at my own pace, in my own words, with a script keeping me on track.

Every teleprompter I tried missed it. They scroll at a fixed speed you have to chase, or they react to sound, not to the actual words you say. Testing video editors and screen recorders, I kept hitting the same gap: plenty show a speaker script, almost none pair it with usable speech recognition.

So I asked: why should I follow the machine, when the machine can follow me? That idea sat in my head for two years. The only way to get rid of it was to build it.

That's Speechius — (probably) the first reliable ASR-based teleprompter. It listens to what you're actually saying and scrolls your script to match — your pace, your words. It also stays hidden from screen recording and screen-share, so the script never ends up in your video.

It's my first app, built in Flutter. Fully on-device (private), one-time purchase, no subscription. macOS & Windows. Founding price for the first 100 buyers: Pro €30 / Studio €60 (then €50 / €100).

👉 speechius.com

First launch — I'd genuinely love your honest feedback, positive or negative. 🙏