How nthoughts protects your voice and your data
Privacy is a pedagogical requirement, not just a compliance obligation. If you suspect your voice or speaking errors might be recorded, sold, or judged, you will self-censor — and self-censorship kills learning. Our architecture is built from first principles to make genuine privacy the default.
No raw audio stored — ever
Your microphone input is transcribed in real time by the browser's WebSpeech API. The resulting text is what we process — not the audio. For pronunciation scoring, a short clip (2–8 seconds) is sent to our AI provider over an encrypted connection for evaluation. It is not retained after the API response. We never use your voice to train AI models.
Mistake logs are encrypted and user-controlled
When Maya identifies an error, the error pattern (not audio) is stored in an encrypted, user-specific database partition. This is what populates your Leitner spaced repetition decks. You can view, export as CSV, or permanently delete this data at any time from account settings.
Session metadata is anonymised
We collect anonymised usage metrics — session count, session duration, features used — for product improvement. This data is never linked to individual user accounts and is not sold or shared with advertising networks.
Authentication via Google OAuth
We use Google OAuth 2.0 for authentication. We receive your name, email, and avatar from Google — nothing else. We do not have access to your Google password, contacts, or other Google data.
Full account deletion
You can delete your entire account and all associated data from account settings. This permanently deletes mistake logs, session records, and preferences. No data is retained after the next backup cycle (maximum 30 days). To request deletion without deleting your account, email care@nthoughts.com.
GDPR compliance
nthoughts supports all GDPR rights: access, rectification, erasure, portability, and the right to object. To exercise any right, email care@nthoughts.com. We respond within 30 days.
Security & Data Handling FAQ
Do you ever store raw audio recordings of my voice?
No. Your spoken audio is processed in real time by browser WebSpeech transcription and ephemeral inference streams. We do not store, archive, or serialize raw audio files onto server disks.
Does nthoughts use my voice data to train public foundation models?
Never. Your practice conversations, transcripts, and mistake logs are never shared with public model training pools or third-party AI brokers.
Where are my personal mistake decks and Leitner cards stored?
Mistake logs are encrypted both in transit (TLS 1.3) and at rest (AES-256) inside your private database partition. Only text representation of grammatical corrections is retained to fuel your personalized Leitner review deck.
How can I export or permanently delete my practice history?
You can export all your mistake cards and performance analytics as CSV anytime from Settings. You can also trigger one-click permanent account deletion, wiping all database records within 30 days.
Is nthoughts GDPR compliant?
Yes. nthoughts fully adheres to European GDPR standards, including the Right of Access, Right to Rectification, Right to Erasure, and Data Portability.
Contact
For support, data deletion requests, privacy questions, or security disclosures: care@nthoughts.com
Full legal detail: Privacy Policy · Terms of Service