Four words today. You learn them, then you say them to her, out loud and face to face. That is the part that works.
About ten minutes. No score, nothing to keep up.
Her turn — for when she is talking
Show a card, she says the word. Tap what happened. Nothing here is a test.
A word is hers after she says it on three different days. Scheduling is FSRS, the same spaced-repetition algorithm Anki uses.
Your voice beats any robot
Record yourself saying a word and the app plays you instead of the synthetic voice, in both modes.
🔒 Recordings stay on this device. They are never uploaded anywhere.
Ayah Polyglot
Follows your phone unless you pick one. Dark is a warm charcoal, for bedtime.
How many unseen words practice mode introduces per language each day. Spaced repetition does the rest: words she misses come back sooner, words she owns drift further apart.
A name helps you tell devices apart in the weekly progress read. Progress backs up to your own Cloudflare space; recordings never leave this device.
Everything lives on this device — progress in local storage, recordings in the browser database. No account, no server, nothing sent anywhere.
Babies learn the sounds of a language from live people, not from speakers — recorded audio alone does close to nothing at this age. So the default mode teaches the grown-up, and the grown-up teaches Ayah.
taza
Copied — paste it wherever the nanny will see it.