Cadence Captions - Transcribe

Cadence is live captioning for deaf and hard-of-hearing users, built so the words land while they still matter.

Everything runs on your iPhone. That is what makes it quick, and it is why your audio and captions never touch a network. No cloud, no account, no telemetry. Cadence is used in medical, legal, and therapy settings, and there is no exception to that rule.


CAPTIONS THE MOMENT YOU OPEN IT

After a one-time permission prompt, the capture screen starts listening on its own every time you open it. No configuration to get through first. Pause, resume, and finish are one tap each, and a status dot always tells you whether it is actually listening.


CAPTIONS THAT FOLLOW YOU INTO OTHER APPS

Switch to another app and captioning keeps going, with the latest line floating in a small window you can move anywhere on screen. Read along while you text, look something up, or take notes, then tap the window to jump back. On by default. Turn it off if you would rather switching apps pause.


FLIP THE SCREEN FOR THE PERSON ACROSS FROM YOU

One tap turns the captions 180 degrees, so someone sitting opposite can read along without leaning over the table. It unflips itself when you stop.


SPEAKER LABELS

On the Default model, see who said what, marked as Speaker A, B, C, with a coloured gutter bar that stays put for the whole conversation. Labels are anonymous and reset every session. No voiceprint is stored, so Cadence never learns who anyone is.


BUILT FOR HOW YOU READ

Scale caption text up to more than twice the standard size, on top of whatever Dynamic Type setting you already use. Choose the typeface captions are drawn in: the system font, Atkinson Hyperlegible, Literata, or OpenDyslexic. Pick light or dark independently of your system setting, for rooms where the phone is the brightest thing on the table. VoiceOver reads each caption as a single element, speaker and text together.


SCROLL BACK WITHOUT LOSING THE THREAD

Scroll up to re-read something and new captions stop yanking you back down. A jump to latest pill brings you back the moment you want it.


EVERY SESSION SAVED ON THE PHONE

Conversations are kept on the phone, newest first, with the date, length, and a preview of the opening line. Tap to read the whole thing back, swipe to delete it for good, and share any transcript as plain text through the normal share sheet with speaker names attached. Only committed text is saved. The audio is never written to disk.


TUNE IT TO THE ROOM

Tell Cadence whether you are talking with a group or one person, and how readily to treat quiet or distant sound as speech worth captioning. Choose how long a pause has to be before a line locks in: snappier for quick back-and-forth, more relaxed for speakers who trail off. A noisy cafe and a one-on-one across a desk are not the same problem.


IT TELLS YOU WHEN SOMETHING IS WRONG

If the captions stall, you get a loud banner and a retry button. If the mic is denied, the banner says so and points you to Settings. Never a frozen screen that looks like nobody is speaking.


CHOOSE YOUR CAPTIONING MODEL

Default leads with accuracy and speaker labels. Alternate trades those for the lowest latency, for when captions are backing up hearing you already have. Both download once and then run offline for good. Built In uses Apple's own on-device speech recognition with no download at all, and needs iOS 26; below that, Cadence falls back to Built In (Legacy).


REQUIREMENTS

iPhone 14 or later running iOS 17 or later. Once a model is on the device, it never needs a connection again.