Live captioning for deaf and hard-of-hearing users. Recognition and speaker labels run entirely on your iPhone. Audio and captions never touch a network.
Hey Product Hunt 👋
I built Cadence because live captioning has a timing problem. Most apps send your audio to a server, wait, and hand you the words a beat too late - by the time you've read the line, the room has moved on. Captions that arrive late aren't captions, they're a transcript.
Cadence runs entirely on your iPhone. Nothing is uploaded, ever - no audio, no transcripts, no account. That's a privacy decision, but honestly it started as a speed decision: skipping the round trip is what makes the words land while they still matter.
A few things I sweated over, all aimed at the awkward moments rather than the demo-friendly ones:
- Captions follow you into other apps. Switch away and the latest line floats in a small window you can move anywhere, so you can text or take notes without losing the thread.
- Flip the screen 180° so the person across the table can read along without leaning over it.
- Speaker labels (A, B, C) with a colored gutter bar - anonymous, no voiceprint stored, reset every session.
- Fonts that read the way you read - system, Atkinson Hyperlegible, Literata, or OpenDyslexic - and text scaling well past Dynamic Type's ceiling.
- Scroll back without getting yanked to the bottom, with a "jump to latest" pill when you're ready.
- Mic tuning for the actual room - a noisy café and a one-on-one across a desk are not the same problem.
- It tells you when it's broken. If captions stall or the mic is denied, you get a banner and a retry - never a frozen screen that just looks like silence.
Sessions are saved on the device, newest first, and you can share any transcript as plain text - useful after an appointment.
I'd especially love feedback from deaf and hard-of-hearing folks on where it falls short: which rooms break it, which settings you reached for first, and what's missing. Fire away in the comments - I'll be here all day.