日本語 · English

Summary

WhisperMemo (the "App") is designed with privacy as its top priority. The App developer does not collect, transmit, or track any personal data. Audio recording, transcription, and AI analysis all happen entirely on your device.

Data we collect

The App developer does not collect any personal data, including:

• Audio recordings
• Transcribed text
• AI-generated titles, summaries, emotions, topics, or key points
• Usage analytics or crash reports
• Location data
• Contacts, calendar, or other device information
• Advertising identifiers (IDFA)

Where data is stored and how long it is kept

Memo content (transcripts, AI summaries, emotion / topic metadata, etc.) is stored inside your iPhone's app sandbox. Where iCloud is available, it is also synced across devices signed into the same Apple ID through Apple's CloudKit Private Database (via SwiftData), entirely within your own iCloud account. When iCloud is unavailable, the App automatically falls back to local-only storage on the device.

Audio recording files are managed according to the retention period you choose in Settings (default: 30 days; alternatives: delete immediately after transcription, or keep forever). If you optionally enable iCloud Drive audio backup, copies are placed inside your private space in Apple's iCloud Drive.

If you enable Obsidian Vault integration, Markdown files are written to the folder you choose. These files live on your device and in any cloud-sync service you have personally configured for that folder.

The Home Screen widget reads memo counts and recent titles from on-device shared storage (App Groups). This information is shared only between the App and its widget on the same device and is never sent off-device.

Neither the App developer nor any other third party can access any of this data.

Device permissions

The App may request the following permissions. Each has a single, clearly stated purpose and can be disabled at any time in iOS Settings.

• Microphone: to record voice memos
• Speech Recognition: to display live transcription while recording (after you stop, the App re-transcribes on-device with WhisperKit)
• Notifications: optional daily morning reminder (scheduled locally, never sent externally)
• iCloud: device-to-device memo sync and optional audio backup
• Files: writing to the folder you select if you enable Obsidian Vault integration

Network use

The App connects to the internet only for the following limited purposes:

• Downloading the speech recognition model (Whisper via WhisperKit) on first launch (host: Hugging Face)
• Optional and only triggered by an explicit user action: downloading the local AI model Qwen 2.5 3B Instruct (~2 GB, host: Hugging Face). This download starts only when you tap the button in Settings.
• iCloud sync and optional iCloud Drive audio backup (only if you enable them, host: Apple)

None of these requests contain your recordings or any personally identifiable information. Model downloads only retrieve public files from Hugging Face using anonymous access (no account).

Speech recognition and AI processing

Speech recognition and AI features (summary, title, emotion, topic) all run entirely on your device. No cloud AI APIs (OpenAI, Google, Anthropic, etc.) are used.

• Speech recognition: WhisperKit (Apple Silicon implementation of OpenAI Whisper), on-device
• AI features (supported devices, default): Apple Intelligence (Apple's on-device LLM via the FoundationModels framework)
• AI features (optional fallback): Qwen 2.5 3B Instruct, run on-device via MLX. On unsupported devices, this is enabled only when you explicitly opt in to download the model from Settings.

Third-party services

The App contains no advertising networks, analytics services (Firebase Analytics, Google Analytics, etc.), or tracking SDKs (Facebook SDK, etc.).

The only third-party services the App communicates with are:

• Apple iCloud: used for syncing and backing up data within your own iCloud account. Governed by Apple's Privacy Policy.
• Hugging Face: used to download public machine learning model files. Accessed anonymously without an account; no user information is sent.

Children's privacy

The App is not directed at children under 13 and does not engage in any child-directed data collection activity of the type contemplated by the U.S. COPPA (Children's Online Privacy Protection Act). Because the App collects no personal data, users of all ages can use it safely.

Governing law

This policy is interpreted and operated in accordance with the laws of Japan, where the App developer is based.

If you use the App from outside Japan, the mandatory consumer-protection or data-protection laws of your jurisdiction (for example, the GDPR in the EU/EEA, the UK GDPR, or the CCPA/CPRA in California) continue to apply to you and are not overridden by this policy. Because the App does not collect personal data, most rights granted by these laws (such as access, correction, deletion, or portability requests) have no corresponding data held by the developer to act upon.

Changes to this policy

If this policy is updated, the "Last updated" date at the top will change. For significant changes, we will also notify users via in-app messaging or release notes.

Contact

For questions about this policy or about App privacy, please contact us at:

nonaka.develop@gmail.com