Privacy
This page describes how AudioReader handles data today. It matches the in-app data-practice copy in Settings → About. AudioReader does not independently certify vendor retention or training terms.
Local-first library
AudioReader keeps imported audiobooks, EPUBs, covers, and on-device transcription on this device. Microphone audio for dictation and shadowing is processed on this device and is not saved as a recording.
Playback, chapter timing, and sentence clip replay stay local. Native dictionary lookup stays local. Learning records live in a local SQLite store. Photos, contacts, location, and advertising identifiers are unused. This app does not use App Tracking Transparency or an advertising identifier.
Optional account
An optional AudioReader account uses first-party email sign-in (email OTP). There is no native social login in this release. Account email and an app-generated device binding id are used for app functionality. Device bootstrap may also send hardware model, hardware name, OS name, OS version, and build number.
Optional sync
When Sync learning data is on, vocabulary, review history, and progress copy to your account. Original audiobook, EPUB, and cover files stay on this device. Audiobook, EPUB, cover, and Anki clip media stay local.
Transcripts
Transcripts copy only when all of the following are true: you grant Copy transcripts to this account, you are signed in with sync on, and the service offers that feature. The operator flag alone is not enough. When upload is off, other devices transcribe locally.
Optional AI
AI is optional. Versioned permission names Alibaba Cloud Model Studio (Qwen, Singapore region) as the processing recipient. When you allow AI, selected text, neighbouring sentences, and book metadata leave this device through AudioReader’s product API. A cache lookup still sends this reading content off the device.
AI reuse is account-private. Retention and training terms are published by Alibaba Cloud Model Studio. This app does not independently certify those terms.
Optional analytics and essential logs
Optional product analytics defaults off. When enabled, aggregate counts and timestamps may be recorded; they do not include reading text, transcripts, saved words, translations, notes, prompts, or audio.
Essential auth, deletion, export, sync-failure, and security product events may still record when you are signed in. Those logs do not include reading text, prompts, OTP codes, or tokens.
Data that can leave the device
When you use the optional account, sync, transcripts, AI, or analytics, AudioReader may transmit: email address; account user id; app-generated device id and device hardware, OS, and build descriptors; other user content (synced learning records, optional transcripts, and AI selected text, neighbouring sentences, and book metadata); product interaction (progress when learning sync is on, and optional analytics events); and other diagnostic data (essential operational events).
These are linked to your account for app functionality, and for analytics only when you opt in. They are not used for tracking as Apple defines it. A signed-in account, optional sync, optional transcripts, optional AI, optional analytics, and essential operational logs can transmit data off the device.
On-device APIs
The app reads and writes app-only preferences. EPUB import checks available disk space before writing extracted files. File timestamps, system boot time, and active keyboards are unused.
Export and deletion
You can export account JSON and request account deletion from Settings → Account. Deletion is requested, then completed by a server job; backups are not instant. Local books remain until you remove them from the library. Removing a book does not delete original files outside app-managed storage.
Contact
Use Settings → Account in the app first. For published-page questions, email audio.reader.service@gmail.com.