Privacy Policy
surviveTheTalk is a language-practice app where you hold spoken conversations with simulated characters to rehearse difficult real-life situations in English. This Privacy Policy explains, in plain language, exactly what data the app collects, how your voice is handled, who we share data with, and the rights you have over your data. It is written to describe what the running software actually does.
Who is responsible for your data
The data controller is Walid Guetarni, the individual operator of surviveTheTalk. For any privacy question or request, contact: guetarni.walid@gmail.com.
What data we collect
- Email address — used to sign you in and to send your one-time login codes.
- Call metadata — for each practice call: the scenario you chose, start time and duration, how many checkpoints you passed, your survival percentage, and the subscription tier you held at the time of the call.
- Debrief summary — after a call, an AI generates a short written summary of how the conversation went (corrections, hesitations, areas to work on). This distilled summary is stored. The full word-for-word conversation transcript is not stored.
- Progress — per scenario, your best score and the number of attempts.
- Subscription records — if you subscribe, a store verification artifact (a signed receipt / purchase token from Apple or Google), the product, the validation status, and the expiry date.
We do not run any third-party analytics or crash-reporting service today, and we do not collect advertising identifiers.
How your voice is handled
During a call, your microphone audio is streamed in real time to a speech-to-text provider so your words can be transcribed and the character can respond. This audio is processed and discarded:
- No recording of your voice (no
.wav, no.mp3) is ever written to disk or to our database. - No voiceprint, voice signature, or other biometric identifier is derived from your voice or retained.
- The text transcript of a call exists only temporarily while the call is in progress; it is used once to generate your debrief summary and is then discarded. Only the distilled debrief summary is kept.
How sign-in works
Sign-in is passwordless. You enter your email, we send a 6-digit code, and you enter it to receive a session token. We do not store passwords.
Third parties we share data with (sub-processors)
To run the service, we send the minimum necessary data to these providers:
| Provider | Purpose | What they receive |
|---|---|---|
| Soniox | Speech-to-text (transcription) | Your live voice audio during a call |
| Groq | Character replies, checkpoint judging, debrief generation | Conversation text (no audio) |
| Cartesia (ElevenLabs as a configurable fallback) | Text-to-speech (the character's voice) | The character's reply text (no audio of you) |
| LiveKit | Real-time audio transport (WebRTC) | The live audio stream between your device and the server |
| Resend | Sending login-code emails | Your email address and the 6-digit code |
| Apple App Store / Google Play | Subscription purchase & validation | The purchase token / receipt (they, not we, handle card & payment data) |
Where your data is stored and how it is protected
Your data is stored in a database on a server located in the European Union (Hetzner, Germany — Falkenstein / Nuremberg). Access to the server is restricted. Data is encrypted in transit (TLS) between your device and our server, and between our server and the providers above.
Your rights
You can delete your account at any time from inside the app (Account → Delete my account). Deleting your account permanently removes your email, your call history, your debrief summaries, your progress, and your subscription records from our database, in one operation. You can also request a copy of the data we hold about you (data export).
If you are in the EU/EEA, you have rights under the GDPR to access, correct, export, and erase your personal data. The in-app deletion and export cover erasure (Article 17) and portability (Article 20); for anything else, email guetarni.walid@gmail.com.
Children
surviveTheTalk is intended for users aged 13 and over (Apple) / PEGI 12 (Google). It is not directed at children below those ages.
Changes to this policy
If this policy changes materially, we will update the effective date above and publish the revised version at this URL.