Privacy Policy

VoiceWiki (“the App”) is a personal voice-wiki app that turns what you say into an automatically organized, private knowledge base. It is built to run on your device. We do not operate servers for your content, we do not use analytics, telemetry, advertising SDKs, or third-party trackers, and we do not require an account. In short: your notes stay on your device.

We collect no personal data. The App has no accounts, no analytics, and no tracking.

The following is created and kept only on your iPhone, inside the App's sandbox:

• Transcriptsof what you say (the text), and the wiki pages, sections, sources, and chat generated from them — stored locally with Apple's on-device database (SwiftData).
• App preferences (e.g., onboarding state, transcription language, AI engine selection) stored in iOS UserDefaults.

Audio is not saved. Your microphone audio is processed live for transcription and then discarded — the App never writes audio files to storage.

The App requests two iOS permissions, used only for the features below:

• Microphone — to capture your voice notes.
• Speech Recognition — to convert speech to text.

You can revoke either permission anytime in Settings → Privacy & Security. If you decline, the App still works — you can type notes instead of speaking.

The App uses Apple's Speech framework to transcribe your voice:

• When your language supports on-device recognition on your device, transcription happens entirely on-device and your audio never leaves your iPhone. The App explicitly requests on-device recognition in this case.
• If your language or device does not support on-device recognition, Apple's Speech Recognition service may process the audio to produce a transcript. In that case, handling of that audio is governed by Apple's Privacy Policy. VoiceWiki itself still does not store the audio or send it anywhere else.

The App makes essentially one kind of network request:

• A one-time download of the on-device AI model (~2.5 GB) from a dedicated, public file-hosting endpoint (model.daige.st), fetched the first time you launch the App (or after reinstalling). No personal data is sent in this request — it only downloads the model file.

Apart from the speech-recognition fallback described in §4 and this model download, the App makes no other network calls. Once the model is installed, organizing and asking your notes works fully offline.

VoiceWiki organizes and answers using AI that runs locally on your iPhone:

• The open-source Google LiteRT-LMruntime executes the downloaded model on your device's hardware. It is bundled in the App and contacts no external service.
• On supported devices, the App may instead use Apple's on-device Foundation Models. This also runs locally and sends nothing to external servers.

Your notes are never used to train any model.

• All your content lives in the App's sandbox on your device. You can delete individual sources, sections, or documents inside the App at any time.
• Deleting the App removes all of its data from your device.
• If you use iCloud/iTunes device backups, the App's on-device data may be included in your own backup, which is controlled entirely by your iOS settings — not by us.

The App is not directed at children under 13, and we do not knowingly collect any personal information from children.

If we make material changes to this policy, we will update the “Last updated” date above and post the revised version at this URL.

Questions, feedback, or data requests: wer2774@gmail.com

Effective: June 6, 2026