Skip to content
Wunder TutorOpen the app

Privacy notice · Beta

What happens to your child’s voice and data

Last updated 18 September 2026. This describes how the Wunder Tutor beta works today. It is written in plain language on purpose; a formal policy will replace it before any public launch.

The short version

  • A grown-up sets the app up and agrees to microphone use. Children use a nickname — we never ask for a child’s real name, email, photo or location.
  • There are no accounts yet. Profiles, scores and progress are stored in the browser on the device you use, not on our servers.
  • To score pronunciation, short voice clips are sent to Microsoft Azure’s speech service. We do not keep those clips on any server.
  • No ads. No third-party analytics or trackers, in the app or on this website.

The microphone

The microphone is active only while the mic button is red. Your child taps to start; recording stops by itself when they finish, or when they tap again, and the microphone is released after every try. If they leave the screen mid-recording, that recording is discarded.

Voice recordings

For scoring: each clip (usually one to four seconds) is sent over an encrypted connection through our server to Microsoft Azure Speech, in the East Asia region, together with the word or sentence your child was asked to say. Azure returns scores for each word and sound. Our server passes the audio along and does not store it.

For replay: so your child can hear “before” and “now”, the newest three recordings of each phrase are kept on the device, separately from the scores. In the Parent Zone you can turn this off — audio is then discarded right after scoring — or delete all recordings at any time.

The teacher’s voice

The model pronunciation is generated by Google’s Gemini service from lesson text only — for example the words “three red apples”. Your child’s voice is never sent to Google. Generated audio is cached so each phrase is made once.

What is stored, and where

  • On your device: nickname, avatar, age, home language, chosen accent, lesson progress, pronunciation scores, the list of sounds that need practice, and (if enabled) replay recordings.
  • On our servers: cached teacher audio for lesson phrases. Nothing about your child.
  • Cloudflare, which hosts the app and this site, processes standard request data such as IP addresses to deliver pages and protect against abuse.

Because data lives on the device, clearing the browser’s site data — or using a different device — starts a fresh profile.

Your controls

Everything is in the app under Me → Parent Zone, behind a grown-ups-only check: turn replay recordings on or off; delete recordings; delete the pronunciation history; delete a learner’s profile; or delete everything.

Conversation practice

In the beta, “Speak with Pip” conversations are scripted: your child chooses from suggested replies and says them aloud. There is no open chat with an AI model and no contact with other people.

Changes and contact

When accounts, syncing across devices or payments are added, this notice will change first, and a grown-up will be asked to agree again. Questions or requests: [email protected].