The fastest way to reach the developer:
Please include your iOS version and a brief description of what you tried. Screenshots help a lot.
MochiBook converts your voice into structured expense entries. The microphone is used only when you press and hold the mic button — no always-on listening. The recording is sent to a transcription service, turned into text, and then discarded.
When you hold the mic, the audio is sent over HTTPS to our Cloudflare Worker proxy, which forwards it to OpenAI's Whisper (for speech-to-text) and GPT-4o-mini (for parsing into amount + category + note). The audio is discarded after transcription; only the structured text entry is saved on your device. The fourth onboarding slide explains this in plain language and you tap Get Started to acknowledge it before any recording can happen. Full details in the Privacy Policy.
Locally on your device, in the app's private container. Daily JSON snapshots back up to the same container automatically. You can export to CSV or JSON from Settings → Data & Backups any time.
No. MochiBook has no user accounts. Apple's App Attest framework authenticates your device to our proxy using a key generated in your iPhone's Secure Enclave — we never see your Apple ID, email, or name.
iOS clears an app's local data when it is uninstalled. Before uninstalling, export your data via Settings → Data & Backups → Back up now (saves a JSON file to your share sheet — keep it in iCloud Drive). After reinstalling, use Import to restore.
The transcription server runs in the US. Round-trip latency is highest from Asia / EU. MochiBook also includes a local fast-path that handles short terse entries ("lunch 200", "拿鐵一百五") without hitting the server at all — these should feel instant. Longer or unusual phrases still go to the AI.
Subscriptions ship in a future update; the current version is free to use with a daily voice cap. When subscriptions are available you'll be able to manage and cancel them via iPhone Settings → your Apple ID → Subscriptions.
See the Privacy Policy for details on data handling, and the Terms of Use for the service agreement.
最快聯絡開發者的方式:
麻煩附上你的 iOS 版本與簡單描述。截圖很有幫助。
MochiBook 把你說的話轉成記帳資料。麥克風只有在你按住 mic 按鈕時啟用,不會背景錄音。錄音上傳轉錄成文字後,音檔即從伺服器丟棄。
按住麥克風時,錄音透過 HTTPS 傳給我們的 Cloudflare Worker 代理,再由代理轉送給 OpenAI Whisper(做語音辨識)與 GPT-4o-mini(把文字解析成金額/分類/備註)。音檔轉成文字後即丟棄,只有結構化的記帳項目存在你的裝置上。第四張 onboarding 投影片會清楚說明這件事,你點「開始使用」即代表同意,之後麥克風才可以送出資料。完整內容見隱私權政策。
存在你 iPhone 上的 app 私有容器。每天會自動產生 JSON 快照備份在同一個容器內。可以隨時從「設定 → 資料與備份」匯出 CSV 或 JSON。
不用。MochiBook 沒有帳號系統。Apple App Attest 會用你 iPhone Secure Enclave 裡的金鑰跟我們的 proxy 認證——我們不會看到你的 Apple ID、信箱、或姓名。
iOS 在你解除安裝 app 時會清掉本地資料。解除安裝前,從「設定 → 資料與備份 → 立即備份」把 JSON 存到 iCloud Drive。重灌後用「匯入」還原。
轉錄伺服器在美國,亞洲、歐洲的網路 round-trip 比較高。MochiBook 有本地快取路徑,簡短的記帳(像「午餐 200」或「拿鐵一百五」)完全不打伺服器,瞬間完成。其他複雜句仍走 AI。
訂閱功能在未來版本提供;目前版本免費可用,有每日語音次數上限。等訂閱推出後,可以從 iPhone「設定 → 你的 Apple ID → 訂閱項目」管理或取消。