Effective October 17, 2026

Privacy Policy

Phonemify collects the minimum data needed to run your personal English pronunciation coach. This page explains what we store, how your voice recordings are processed, and who receives data when you practice.

1. How voice recordings are handled

• Audio recordings are not saved to our database or disk. When you record a sentence, the audio is held in memory only long enough to score your pronunciation, then discarded immediately.
• Third-party speech & AI processing (Azure and Gemini). To score your pronunciation and generate coaching feedback, your audio recording and target sentence are sent to Microsoft Azure (Azure AI Speech Pronunciation Assessment and Neural Text-to-Speech) and/or Google Gemini (for holistic speech coaching, personalized tips, and generating practice sentences from pasted text).

2. What we store

If you sign in with Google

• Google account profile: your stable Google user ID (sub), email address, display name, and profile picture URL from Google's OpenID Connect ID token.
• Your practice history: for each practice take you complete while signed in, we store the target sentence, the text transcript of what the engine heard, your overall and phoneme-level scores, problem sounds, and coaching tips so you can track your progress across devices.
• Monthly usage count: a counter of how many recordings your account has analyzed in the current calendar month.

If you use the free trial while signed out

• To enforce the anonymous free-trial limit, we store a monthly counter keyed on a random browser cookie (voice_uid), a salted SHA-256 hash of your IP address, and a salted hash of standard browser headers (User-Agent and language). We never store raw IP addresses.
• Anonymous trial attempts are not linked to a personal profile and cannot be viewed in the history feed.

3. Who we share data with

• Google (Sign-In & Gemini API): When you click "Sign in with Google", you authenticate directly with Google. When coaching tips, free-speak analysis, or custom practice sentences are generated, the relevant audio or text is sent to the Google Gemini API.
• Microsoft Azure (Speech Services): Your recorded audio clip and target phrase are sent to Microsoft Azure Speech Services to compute word- and phoneme-level pronunciation scores and synthesize model audio.
• Railway: Our hosting infrastructure provider, where our application server and SQLite database run.
• Brevo (optional feedback form): If you submit a message through the feedback box, your message (and optional reply email) is delivered to us via Brevo's transactional email API.

We do not sell your personal information or use advertising trackers.

4. Cookies

• session — set when you sign in with Google. Stores your basic identity claims in an HttpOnly, SameSite=Lax signed cookie so you stay signed in across page loads.
• voice_uid — an HttpOnly, SameSite=Lax random identifier used to count anonymous free-trial takes before sign-in.

5. Your rights & data deletion

You can request a copy of your stored practice history or ask us to delete your account and all associated practice attempts at any time by using the feedback form on the home page or emailing d.duque25@gmail.com.

6. Changes & contact

If we make material updates to this policy, we will update the effective date at the top of this page. See also our Terms of Service. For any privacy questions, contact d.duque25@gmail.com.