Wen Privacy Policy

Last updated: 2026-08-23

Wen (闻文) is an iOS app for AI narration of novels you import yourself. Our principle: collect as little as possible, and keep it on-device wherever we can.

What we collect

• Imported text: Book files (txt/epub) you import are stored locally on your device. We do not upload or archive them on our servers, and never use them for model training.
• Narration requests: The free on-device tier synthesizes speech entirely on your device — nothing is sent to our servers. Paid cloud tiers (Standard/Premium/Flagship) send the sentence being read to the corresponding voice provider (listed below) to generate audio.
• On-device semantic search: Library search builds its index locally using Apple's NLContextualEmbedding. Your text never leaves the device for search.
• Usage & credits: We log narration usage (character counts, timestamps) to run the credits system — never the narrated content itself.
• Purchases: In-app credit purchases are handled entirely by Apple's StoreKit. We never see or store your payment details.

Who we share data with

Only when you choose a cloud narration tier does the text being read leave your device, for the following purposes and nothing else:

• Speech synthesis: the sentence being read is sent to your tier's voice provider — Microsoft Azure Speech (Standard), MiniMax (Premium) or ElevenLabs (Flagship) — to generate audio in real time.
• Character & tone analysis (Theatrical mode only): a short window of the passage being read (never the whole book) is forwarded via OpenRouter to large-language-model providers to decide who is speaking and how. We currently use models from Anthropic, DeepSeek, Google and OpenAI; the exact models may change with quality and cost, always limited to this purpose.
• Smart Import (optional): when you choose Smart Import, the catalogue table's headers, a few sample rows and the file names (no book text) are sent the same way to identify the title and author columns.

We ask for your explicit consent before the first time any cloud tier is used. You can switch back to on-device narration at any time to keep your text entirely on your device.

To avoid re-processing identical content: generated audio may be cached briefly in our storage (Cloudflare R2), and Theatrical-mode analysis results (character rosters, tone tags) are cached keyed by a hash of the text. Both are content-keyed, tied to no account, and used for nothing else.

What we don't do

• We do not host or provide any book content — you may only import text you already own
• No ad tracking, no third-party advertising SDKs
• We do not sell your data to anyone

Deleting your data

You can delete any imported book and its local data at any time within the app. You can also delete your account and its associated cloud data (usage and purchase records) in Settings → Account → Delete account — no need to contact us.

Age rating

The app itself hosts and provides no book content — everything is imported by you. We do not knowingly collect personal information from children.

Contact

Questions about this policy or your data: thomasxinhaili@gmail.com.

Changes

We'll notify you in-app of any material changes to this policy.