Privacy Policy

Controller: [LEGAL ENTITY NAME], [REGISTERED ADDRESS] ("Yapr," "we," "us")

Version history: v3.2 — 16 Jul 2026 (new product surfaces, retention defaults filled, Cloudflare + Resend processors) · v3.1 — 12 Jul 2026 · v2 — 8 Jul 2026 · v1 — 21 May 2026.

1. The short version

Yapr translates conversations in real time and can speak the translation in a clone of your voice. Your voice is one of the most personal things you have, so our position is simple:

• We never store your conversation audio or transcripts on our servers.
• Strict mode runs entirely on your device — in that mode, nothing leaves your phone and our servers are never contacted.
• You can delete your voice clone and your account at any time, end to end.
• We do not sell your personal information and do not use your voice or conversations to train AI models.

This summary is for convenience; the full policy below governs.

2. Who this applies to & age

Yapr is for adults and is not directed to children. Because creating a voice clone processes biometric data, you must be at least 18 years old to use Yapr, and we ask your age before collecting any personal data. We do not knowingly collect data from anyone under 18; if you believe a minor has provided us data, contact privacy@yap-r.com and we will delete it.

3. The two privacy modes (read this first)

Yapr has a single, always-visible privacy toggle that determines whether a conversation touches the cloud at all.

Standard mode (default)

Best translation and voice quality. Speech recognition, translation, and voice synthesis run through trusted third-party processors (see §6). We are explicit about what is and isn't retained (§5).

Strict mode

True zero-retention. Speech recognition, translation, and playback all run on your device. Our servers and third-party processors are not contacted for the conversation, so there is nothing for us — or anyone — to store. Quality is lower; we state that plainly in the app. Strict mode also powers offline Emergency phrases.

4. Information we collect

You provide:

• Account identity — your email address and/or the identifier from your sign-in provider (e.g., Google). Required to create and secure your account. You sign in either with Google or with a passwordless one-time code (OTP) emailed to you — either way we process your email address to create and secure the account; we never ask for or store a password.
• Security settings (optional) — if you turn on two-factor authentication (2FA), you set up a time-based one-time-code (TOTP) authenticator. The 2FA factor is held by our authentication provider (Supabase Auth); we store only whether 2FA is on for your account.
• Voice enrollment samples — short audio recordings you make to create your voice clone, plus a consent recording in which you state that you consent to cloning your own voice.
• Profile & learning preferences — set during onboarding and in Learn: your target language, your stated learning goal, and course progress (units completed, XP, streak). Used to generate and track your personalized course.
• Settings — target language, privacy-mode preference, per-mode voice preferences.

Collected automatically (Standard mode):

• Service metadata — for each request: timestamps, languages, mode, and request counts/durations (for reliability and billing). No conversation content.
• Device/technical data — app version, OS version, coarse diagnostic data, and crash reports (only if you consent to crash reporting).

Stays on your device (we never receive it):

• "Save phrases from your conversations" (Learn↔Live) — an opt-in, off by default feature that, when you turn it on, saves phrases you said in a Live conversation locally on your phone so you can practice them later in Learn. This capture is entirely on-device; the saved phrases are never sent to our servers and there is no server copy.
• Medical ID, emergency contacts, and precise location (Emergency/SOS) — stored only on your device; see §5.

We do NOT collect: precise location, your contacts, your photos, or advertising identifiers.

5. What we store, and what we never store

We never store (in either mode):

• Conversation audio (server-side).
• Transcripts of your conversations.
• Translation text beyond transient, in-memory processing needed to return the result.
• Your location or contact lists.

We store (Standard mode only — Strict mode stores nothing server-side):

• Account identity (email / sign-in id).
• Voice-clone metadata: the voice provider's voice ID, a label, and creation timestamp — needed to address your voice for playback. (The voice model itself is held by our voice provider; see §6.)
• Your settings.
• Billing-relevant request counts and durations (no content).
• Server logs limited to an allowlist of non-content fields (account id, status, duration); retained ~30 days.

Voice enrollment samples are used only to create your clone and are deleted immediately after the creation attempt completes — whether it succeeds or fails. The consent audio is verified and then deleted in the same step; what we keep is a consent record (the notice version you saw, a transcript, and a timestamp — not the audio), retained for the life of your account as proof of consent. Full detail: our public Biometric Data Retention & Destruction Schedule.

6. Third-party processors (Standard mode)

In Standard mode we share the minimum necessary with processors who act on our instructions under data-processing agreements. We require Data Processing Agreements (DPAs) with each before launch.

In Rooms (two-phone translation), call media travels directly peer-to-peer between the two paired devices. When a direct connection isn't possible across networks, the media is relayed through Cloudflare's TURN service, which passes the end-to-end-encrypted media through without being able to read it and sees only IP/connection metadata; Supabase carries only the brief session-setup (signaling) messages.

Each processor retains data per its own terms; we use no-training / no-retention settings where the provider offers them. In Strict mode none of these are contacted. We do not claim Zero-Retention Mode for Standard-mode providers unless and until we hold the contracts that permit it. The current processor list, locations, and transfer mechanism for each are kept in our public sub-processor register.

7. How we use your information

• To provide translation and voice features you request.
• To create, operate, and (on your request) delete your voice clone.
• To authenticate you and keep your account secure.
• To maintain reliability, prevent abuse, and meter usage for billing.
• To comply with law.

We do not use your voice, recordings, or conversations to train our own or third parties' AI models, and we do not sell or share your personal information for advertising.

8. Legal bases (EU/UK GDPR)

• Contract — to provide the service you sign up for.
• Consent — for creating your voice clone (a special category of data); you give explicit consent via the in-app consent recording, and you may withdraw it at any time by deleting your clone.
• Legitimate interests — security, abuse prevention, and basic service metadata, balanced against your rights.
• Legal obligation — where retention/disclosure is required by law.

9. Your rights & choices

Depending on where you live (EU/UK GDPR, California CCPA/CPRA, Illinois BIPA, and others) you may have the right to: access, correct, delete, port/export, restrict or object to processing, and withdraw consent. Yapr provides:

• Delete your voice clone — in Settings; hard-deletes the clone at our voice provider and our metadata row.
• Delete your account — removes your profile, voice clone, and stored settings.
• Export your data — a machine-readable copy of your profile, settings, and conversation metadata (we hold no transcripts/audio to export).
• Withdraw consent for your voice only — delete your voice clone in Settings without deleting your account; this destroys the voiceprint (including at our voice provider) and you can keep using Yapr with a standard voice. Withdrawing is as easy as consenting was.

To exercise rights, use the in-app controls or email privacy@yap-r.com. We do not sell personal information; this also serves as our "Do Not Sell or Share" statement for CCPA/CPRA. We do not use sensitive personal information beyond the purposes permitted by CPRA §1798.121, so no separate "Limit the Use of My Sensitive Personal Information" link is required. [Counsel confirm.]

EU/UK: you also have the right to lodge a complaint with your supervisory authority (your local DPA, or the UK ICO). Our lead authority is [TBD once EU representative / establishment is set].

Illinois (BIPA) note: your voiceprint is a biometric identifier. We collect it only with your informed written/explicit consent (a dedicated in-app release, captured before any recording), use it solely to provide the voice feature, never sell, lease, trade, or profit from it, and destroy it per our public Biometric Data Retention & Destruction Schedule. [A lawyer must confirm BIPA-specific consent wording, retention schedule, and disclosures.]

Texas (CUBI) & Washington (My Health My Data): in Texas we collect a voiceprint only with notice and consent, don't sell it, and destroy it on our schedule. Washington's My Health My Data Act defines "consumer health data" broadly; to the extent it treats a voiceprint as in scope, we rely on your consent and provide access/deletion via the same channels. [Counsel to confirm whether a voiceprint is "consumer health data" under Washington MHMD and whether a separate MHMD notice/consent is required.]

Canada (PIPEDA / Quebec Law 25): we rely on express consent for biometric data; Quebec residents' rights (access, rectification, de-indexing, portability) can be exercised via the same channels. [Quebec availability follows the CAI biometric-database filing.]

Brazil (LGPD): where available in Brazil, you have the LGPD rights (confirmation, access, correction, anonymization, portability, deletion, revocation of consent) via the same channels; our legal basis for the voiceprint is consent (Art. 11(I)).

Other jurisdictions (Australia, New Zealand, Japan, Korea, Mexico, and others where Yapr is offered): we honor access, correction, and deletion requests through the same in-app controls and email, regardless of where you live.

10. Security

• Encryption in transit (TLS) for all network traffic.
• Provider API keys are held only on our server; they are never shipped in the app.
• Your voice playback endpoint is authenticated — only your signed-in session can trigger your voice.
• Optional two-factor authentication (2FA/TOTP) you can enable in Settings, and a step-up check before sensitive account actions.
• Access controls and audited backend functions; logs exclude conversation content.

No system is perfectly secure; we cannot guarantee absolute security.

11. AI-generated content disclosure (EU AI Act Art. 50)

Yapr's translated speech is AI-generated audio — including, if you opt in, audio generated in a clone of your own voice. To keep synthetic speech honest:

• The app tells both parties when they are hearing an AI-generated voice, and cloned-voice playback is clearly disclosed in the conversation UI.
• Synthesized audio produced by Yapr carries a machine-readable marking identifying it as artificially generated, in line with EU AI Act Article 50. [Engineering in progress — must be live before EU availability; regulation applies from 2 Aug 2026.]
• Your voice clone can speak only through your signed-in session — we don't offer your voice to anyone else, and there is no public voice library.
• Talking to an AI character: some features have you converse with an AI, not a person — the Learn Roleplay practice character (a waiter, hotel clerk, etc.) and the Voice Setup Call assistant that helps you enroll your voice. These use a standard (non-cloned) synthetic voice, and we tell you they're AI (EU AI Act Art. 50(1)).
• Using Yapr to impersonate someone or to present synthetic audio as authentic human speech is prohibited (see Terms).

12. International transfers

Yapr's processors operate primarily in the United States; using Yapr from outside the US means your data (including, if you opt in, your biometric voiceprint) is transferred there. For transfers from the EU/EEA, UK, and Switzerland we rely on, in order of preference:

1. the EU–US Data Privacy Framework (and its UK Extension / Swiss counterpart) for processors holding an active DPF certification — our voice provider ElevenLabs is DPF-certified [reverify "Active" at signing];
2. Standard Contractual Clauses (SCCs) plus a documented Transfer Impact Assessment for processors without DPF certification;
3. [If Supabase is moved to an EEA region, account/metadata storage ceases to be a transfer at all — decision pending.]

The current processor list, locations, and transfer mechanism for each are kept in our sub-processor register. [Counsel: confirm mechanisms per processor at DPA signing; keep the register and this section in sync.]

13. Data retention

• Conversation audio/transcripts: not retained.
• Voice enrollment samples: deleted immediately after the creation attempt (success or failure).
• Consent audio: deleted after verification; the consent record (notice version, transcript, timestamp): life of your account, deleted with the account.
• Voiceprint: until you delete your voice, withdraw consent, delete your account, or after 24 months of inactivity — whichever comes first; destruction propagates to our voice provider. Full schedule: Biometric Data Schedule.
• Account + metadata: until you delete your account, then removed within 30 days (subject to legal/backup constraints).
• Server logs: ~30 days.

14. Health-data & HIPAA

Yapr is not a HIPAA-covered entity and does not claim HIPAA compliance. For privacy-sensitive conversations we offer Strict mode, where nothing leaves your device. Do not rely on Standard mode for protected health information.

15. Changes

We'll update this policy as the product evolves and post the new effective date. Material changes will be notified in-app or by email.

16. Contact & representatives

Privacy questions: privacy@yap-r.com
Security reports: security@yap-r.com

Controller: [LEGAL ENTITY + REGISTERED ADDRESS]

EU representative (GDPR Art. 27): [NAME + ADDRESS — appoint before EU availability]

UK representative (UK GDPR Art. 27): [NAME + ADDRESS — appoint before UK availability]

Data Protection Officer: [NAME/CONTACT if appointed — likely required; counsel to assess]

See also our Terms of Service, Biometric Data Schedule, Sub-processors, Accessibility statement, and how to delete your account.