言葉スナップ · Kotoba Snap

简体中文

Effective date: 2026-07-28 Last updated: 2026-08-07

Kotoba Snap (“the app”) is a Japanese vocabulary learning tool: you photograph or pick an image, the app recognizes the Japanese text in it, and helps you turn unfamiliar words into flashcards with a review schedule.

This policy explains how the app handles your information. The app has no accounts — you do not register or sign in, and every feature works without doing so.

1. What the developers never receive

We (the developers of this app) operate no server, and therefore never receive or store any of the following:

The app’s own code never transmits any of it anywhere.

That is not the same as “nothing ever leaves your device”. Two paths do: your phone’s own operating-system backup (section 2) and diagnostic data sent by the Google ML Kit SDK this app integrates (section 4). Please read both.

2. Where your data lives

All of the following is stored in the app’s private storage area on your device:

You can cap how much space source images may use, or clear all source images at once, from Settings. Deleting a flashcard also deletes its review history.

Operating-system backup and moving to a new phone

This app provides no cloud sync operated by us, and never sends your images, flashcards, or review data to our servers.

However, your phone’s own operating-system backup may back up and restore some of this app’s data, depending on your system settings. In the current version:

  iPhone (iCloud Backup) Android (Google Auto Backup)
Flashcards, review history, settings Included in system backup by default Included in system backup by default
Source image copies Explicitly excluded from backup Explicitly excluded from backup

“Included by default” describes the platform’s default behaviour. Whether a backup actually exists, and whether it can be restored, depends on whether you have system backup enabled and whether it completed successfully.

Source image copies are explicitly excluded on both platforms, to keep fewer copies of your photos in the cloud or in transit. On Android this exclusion applies to cloud backup, to direct device-to-device transfer when setting up a new phone, and to transfers to a non-Android device.

So after moving to a new phone: your flashcards and review history come back; the source images do not. The “by source image” grouping still exists and still works — it simply shows a placeholder instead of the photo.

These backups live in your own Apple or Google account, which we cannot access. Whether backup is on, and how to delete an existing backup, is governed by your system settings and Apple’s or Google’s policies.

Uninstalling the app deletes all local data on that device. If your system has already made a backup, what happens to that copy is determined by those platform backup settings, not by us.

3. Text recognition happens on your device

Text recognition (OCR) runs locally on your device:

Your images and their recognized text are not sent to us, and are not sent to Google. On the latter, Google states in the ML Kit Terms of Service that when you use ML Kit APIs, processing of the input data (images, video, text) fully happens on-device, and ML Kit does not send that data or the resulting outputs to Google servers. Apple’s VisionKit is a built-in on-device system capability and likewise recognizes text locally.

Japanese word segmentation, dictionary lookup, and review scheduling are likewise entirely local — the dictionary ships with the app and needs no network connection.

3.1 Reference translation is also on-device, but its language pack is downloaded

The app can show a sentence-by-sentence reference translation (Japanese → Simplified Chinese) of the recognized text. Translation also runs locally on your device:

Neither your images nor the recognized text are sent anywhere in order to translate them — not to us, not to Google or Apple.

One part does use the network, and we state it separately: the translation language pack does not ship with the app and has to be downloaded the first time you use the feature. That is a real network request, it happens only after you tap to start it, and what is downloaded is a general-purpose language model — the request carries none of your images or text. Once downloaded the pack stays on your device and translation needs no network afterwards.

On Android that download is performed by Google’s ML Kit; on iOS it is performed by the system and follows your own cellular-data settings.

3.2 What needs the network, and what does not

In one sentence: apart from downloading a language pack the first time you use Reference Translation, nothing in this app needs a network connection.

Feature Needs network?
Taking a photo / picking from the library No
Text recognition (OCR) No
Furigana No
Dictionary lookup, candidates, glosses No (the dictionary ships with the app)
Creating, editing, searching and organising cards No
Review and review scheduling No
Revisiting a source image and its recognized text No
Backup export / import No
Reference translation — first use Yes: downloads a language pack, once
Reference translation — after the pack is downloaded No

In other words: once the pack has been downloaded, the whole app works in airplane mode.

One more thing needs stating separately because it is not tied to any feature: the Google ML Kit SDK integrated in this app automatically sends limited diagnostic data (see section 4). It is not triggered by any particular feature, and it does not include your images or text.

4. Diagnostic data collected by a third-party SDK

The app does use the network, and we cannot claim that no data ever leaves your device.

The Google ML Kit SDK integrated in this app automatically sends limited diagnostic and usage metrics to Google. This version provides no in-app setting to turn that SDK telemetry off.

Per Google’s official data disclosures (Android / iOS), that information includes:

Common to both platforms

Android only, in addition

Google states that this data is transmitted over HTTPS and is not shared with third parties. These diagnostic metrics do not include your images or the recognized text — as described in section 3, Google states this explicitly in the ML Kit Terms of Service.

For how Google handles this data, see Google ML Kit Terms & Privacy and the Google Privacy Policy.

Beyond the ML Kit diagnostic and usage metrics described above, the app contains no advertising SDK, no general-purpose behavioural analytics SDK, and no crash-reporting SDK.

5. Permissions

Permission Why
Camera Used only when you choose “Take photo”, to capture the image to be recognized
Photos / Photo library Used only when you choose “Import from library”. The app receives only the single image you selected; it does not read your library as a whole

The app requests no location, contacts, microphone, phone, or SMS permissions.

On Android this app declares no permissions of its own. The app info page may show INTERNET and ACCESS_NETWORK_STATE; those come from the Google ML Kit components described in section 4.

6. Children

The app is not directed at children under 13. It has no accounts and asks for no name, email address, date of birth, or any other identifying information.

Where images and flashcards are stored, and how operating-system backup treats them, is described in section 2; the Google ML Kit diagnostic data is described in section 4. These apply the same way to everyone, including children.

7. Your rights

We hold none of your content on any server of ours, so your control over that content is entirely on your device:

Two things we cannot do on your behalf:

If you have questions about how the app handles data, or want help understanding any of the above, contact us at the address below.

8. Changes to this policy

If this policy changes, we will update the “Last updated” date at the top of this page. Where required by law or otherwise appropriate, we will provide additional notice.

9. Contact us

kotobasnap.support@gmail.com

We will do our best to respond within a reasonable time.