ListeningDrillAIリスニング特訓

Privacy Policy

Last updated 2026-11-05 · ListeningDrill (AIリスニング特訓)

01What we handle

DataWhenWhere
Exam target, speech rate, reminder flag, weakness tags, drill sets, English passages, questions/answers, archive summariesWhile you study, configure, or archiveOn-device Application Support/ListeningDrill/*.json
Worksheet photo JPEGs (camera)When you build a drill from a photoStored under captures/ on device. Pixels are not uploaded
OCR text from that photoAfter capture, before generationOn-device Apple Vision; only the text may be sent to DeepSeek
Microphone audio and speech transcriptWhen you dictate a practice requestApple SFSpeechRecognizer (ja-JP). Audio may be processed on Apple servers because on-device recognition is not forced
AI consent stateWhen you allow or revokeOn-device ai-consent-state.json
Exported worksheet / score imagesWhen you exportYour Photos library (add-only permission)

No accounts or logins. We do not collect name, email, phone, location, advertising IDs, or face data. No analytics or ad SDKs.

02External AI (DeepSeek)

Only after you enable Allow external AI in Settings, generation and explanations send the following text to DeepSeek (api.deepseek.com, model deepseek-chat):

  • Exam target and default speech rate
  • Weakness tag names
  • Up to about three recent archived AI summaries
  • Your typed instruction or spoken transcript
  • OCR text from a worksheet photo (up to ~1800 characters). The image file itself is not sent
  • For explanations: stems, options, correct index, and your answers

DeepSeek processes requests in China. Responses (passage/question JSON or a Japanese explanation stream) stay on your device. Revoking consent stops further sends.

The in-app consent copy highlights profile, weakness tags, and answers. In practice OCR text and speech transcripts are also included when you use those features. Review content before sending.

03Cross-border processing

With your consent, the learning text listed above is processed outside Japan (China) by DeepSeek. Apple speech recognition may also leave the device depending on your system settings. Processing is limited to what those features need.

04On-device speech and notifications

Listening playback uses on-device AVSpeechSynthesizer only—no external TTS. Reminders are local notifications on your device.

05Retention, deletion, export

  • Data stays on your device. Uninstalling removes the app sandbox.
  • You can revoke AI consent in Settings. There is no in-app “delete everything” control (aside from a developer acceptance reset).
  • You can export worksheet or score images to Photos. For extra deletion help, email sheetbox@inkbench.studio.

06Children

The app is aimed at high-school and exam learners. If you are under 13, use it only with a parent or guardian who can review consent and what is sent.

07Changes

We will update the date on this page when the policy changes. Material changes may also be noted in the app or here.

08Contact

Privacy questions: sheetbox@inkbench.studio