Privacy Policy

How we process your data

Darija to English helps signed-in users understand Moroccan Darija audio and create Arabic/Darija voice replies. This policy explains what data is processed to provide those features.

Effective date: June 2, 2026

Data We Process

We process account information such as email address, user identifiers, profile name, session information, and consent records.

When you use translation or reply features, we process audio recordings, uploaded filenames, transcripts, translations, generated reply text, generated audio, timestamps, job identifiers, provider status, and diagnostic metadata needed to operate and support the app.

How We Use Data

Service Providers

We use Clerk for authentication, Convex for app state, consent proof, private storage, and saved outputs, Trigger.dev for background job orchestration, OpenAI for transcription, translation, and speech generation, and a Modal-hosted Darija speech recognition service for Moroccan Darija accuracy.

Retention

By default, original recordings are temporarily stored only to complete processing, then deleted from app storage after worker processing. Unsaved results expire automatically under our app retention settings.

If you choose to save a result, saved history may retain transcripts, translations, generated audio, filenames, timestamps, and job metadata until you delete your account or request deletion. Consent proof may be retained in anonymized form as compliance evidence.

Our providers may process and retain limited data under their own security, abuse-prevention, logging, backup, or legal requirements.

Your Controls

High-Stakes Content

Do not use Darija to English for legal, medical, emergency, immigration, financial, or other high-stakes decisions. AI-generated outputs may be inaccurate.