Privacy
Privacy policy
Last updated: 10/28/2026
In short
We ask before a recording leaves the device. Your clip is scored by Microsoft Azure Speech and then discarded. We keep structured results (scores and error markers), not your voice. MiAcento is for users 13 and older.
What we collect
When you check pronunciation, and only after you allow it in the app or on the site, microphone audio and the practice phrase are sent to Microsoft Azure Speech. The raw file is not written to disk or saved in our database. On the website you get an anonymous session id (web-xxx); in the app, a stable anonymous UUID on the device. A display name is optional and stays with us. We do not collect date of birth.
Ephemeral audio processing
The clip exists in memory only for the duration of the Azure Speech request. What we retain is a structured result: scores and timestamps of problem sounds — not the raw waveform. We do not keep a library of student recordings. That design reduces biometric-data concerns around stored voice.
Age 13+
The Service is intended for users 13 years of age or older. At app start we ask for a self-declared confirmation that you are 13+ (a checkbox — we do not collect date of birth as a separate field). Under US children's privacy rules, if we obtain "actual knowledge" that a user is under 13, we must stop providing the Service to that user.
Mobile app
The device UUID matches the RevenueCat App User ID so we can restore a subscription. The app asks for the microphone only to score pronunciation. Local practice reminders do not send data to the server. PRO is purchased via the App Store / Google Play; the store and RevenueCat handle payment — we never see your card or billing profile.
Who receives data
Your voice recording and the practice phrase are sent to Microsoft Azure Speech (pronunciation scoring) only after a separate permission in the app or on the site. The phrase text, the scores, and your native language — not the audio, your name, or your email — are sent to OpenRouter and then to the language-model provider to write the teacher’s feedback. The practice-phrase text, not the learner’s voice, may be sent to ElevenLabs to create the teacher’s voice; finished teacher clips are stored by us. Without that permission, the recording is not transmitted. We do not keep a copy of your recording. Payments are handled by Apple, Google, and RevenueCat. Waitlist and news emails are sent by Resend.
Deleting data
In the app (Menu → Delete account) you can erase your profile and progress. There are no stored voice recordings to delete. On the website, clearing browser data is enough — the web-xxx session is no longer linked to new attempts. Questions: .
Contact
Data questions: . See also the contacts page.