RHETORIA
Secure, Local, Offline Transcription & Notes
Rhetoria is a powerful, secure transcription and AI content tool that lives entirely on your device. Whether you are a journalist handling sensitive interviews, a student taking lecture notes, or a professional on the move, Rhetoria ensures your data stays exactly where it belongs:
with you
Features
Privacy Without Compromise
Unlike other services, Rhetoria processes everything locally.
100% Offline: Transcribe and transform audio without an internet connection.
Airplane Mode Ready: Perfect for remote work, flights, or areas with no service.
Zero Data Collection: Your transcripts and audio files never leave your device.
The LOCAL AI Workbench
Turn your voice notes into actionable content instantly. Using our built-in, local AI models, you can generate:
Professional blog posts and articles
Academic papers and summaries
Social media content
Meeting minutes and checklists
All generated locally—no external AI APIs used.
Bank-Grade Security
We take your data protection seriously. Rhetoria uses industry-leading standards to keep your information locked down.
AES-256 Encryption: Your data is encrypted at rest using the most secure standards available.
Biometric Protection: Secure your vault with fingerprints, face recognition, or a custom password.
Local Decryption: Data is only decrypted when you authorize it.
Why Choose Rhetoria?
If you value your privacy but need the power of modern AI, Rhetoria is the solution. Avoid the surveillance and risks associated with live transcription services and cloud-based AI models.
Take control of your data.
Download Rhetoria today.
PRIVACY POLICY
Summary
Rhetoria is designed to keep your voice recordings, transcripts, and generated content on your device. We do not operate servers that receive your content, we do not run analytics, and we do not share data with third parties. This policy explains the limited circumstances in which the app uses the network, what is stored on your device, and the controls available to you.
Who we are
Rhetoria ("the app", "we") is an on-device transcription and AI content tool. References to "your device" mean the Android device where you installed the app.
What data Rhetoria processes
On-device only — never transmitted to us:
Audio recordings captured by the app.
Transcripts produced from those recordings.
AI-generated content (summaries, blog posts, social posts, meeting notes, action items, emails, mind maps, academic drafts) produced from your transcripts.
Vault credentials (password, optional biometric unlock material).
All of the above are stored locally and encrypted at rest (see "Security" below).
Not collected at all:
No account, email, phone number, or sign-up.
No analytics, telemetry, crash reporting, advertising IDs, or usage metrics.
No contacts, location, calendar, photos, or other device data.
When Rhetoria uses the network
Rhetoria requests the INTERNET and ACCESS_NETWORK_STATE permissions for a single purpose:
One-time model download. On first launch, the app downloads the local AI model from Hugging Face (huggingface.co) so that transcription and content generation can run entirely on your device thereafter. The app's network security configuration restricts outbound HTTPS to that host.
After the model is downloaded, the app can be used fully offline, including in airplane mode. Your audio, transcripts, and generated content are never uploaded to Hugging Face, to us, or to any other service.
Hugging Face's handling of the download request (e.g., IP address, request metadata) is governed by Hugging Face's own privacy policy. We do not receive any data from that request.
How transcription and AI generation work
Transcription and content transformation are performed by a local model running on your device via Google's LiteRT runtime. No audio, transcript text, or prompt is sent to an external API, including OpenAI, Anthropic, or Google's cloud services. Generation happens entirely within the app process.
Permissions
Record audio (RECORD_AUDIO) — required to capture recordings.
Internet / network state (INTERNET, ACCESS_NETWORK_STATE) — used only to download the local AI model on first launch and to check network availability for that download.
Biometric (USE_BIOMETRIC / USE_FINGERPRINT) — optional; used to unlock your encrypted vault if you enable biometric unlock.
Security
Encryption at rest. Audio segments are encrypted with AES-256-GCM. The local database is encrypted using SQLCipher with a 256-bit data-encryption key derived from your password using Argon2id.
Authentication. Access to your vault requires your password and, optionally, a strong biometric (fingerprint or face) that wraps the key on-device.
Local decryption. Data is only decrypted in memory, on your device, after you authenticate.
No backups to us. We do not back up your content. If Android system backup is enabled for the app on your device, that is governed by your device and Google account settings, not by us.
Data retention and deletion
Your content remains on your device until you delete it from within the app, clear the app's data in Android settings, or uninstall the app. Because we never receive your content, there is nothing for us to delete on our side.
Children's privacy
Rhetoria is not directed at children under 13. We do not knowingly collect data from anyone, including children.
Your rights
Because Rhetoria does not collect personal data, there is no user data for us to access, export, correct, or erase. You retain full control over the content stored on your device.
Third parties
Hugging Face hosts the AI model file downloaded on first launch. We do not share any user content with Hugging Face; the only request made is a standard HTTPS download of a publicly hosted model.
No advertising networks, analytics providers, or marketing partners are integrated into the app.
Changes to this policy
If we change this policy, we will update the "Last updated" date and publish the new version with the app. Material changes will be highlighted in release notes.