隱私權政策
最後更新:2026 年 8 月 27 日
這份政策說明「水豚學韓文」(以下稱「本 App」)會處理哪些資料、為什麼要處理、 以及資料會去哪裡。我們盡量把實際發生的事寫清楚,而不是寫得含糊但安全。
1. 我們處理哪些資料
1.1 只存在你裝置上的
以下資料存在你的裝置本機,不會離開裝置:
- 學習進度:學過的韓文字母與單字、熟悉度
- 測驗紀錄:測驗次數、答對題數、正確率
- 連續天數:目前連續天數、最長紀錄、最後學習日期
- 成就:已解鎖的成就徽章
- 設定:音效開關、每日目標、韓語程度
1.2 我們不會收集的
- 姓名、電子郵件或其他可識別個人身分的資訊
- 帳號與密碼——本 App 沒有登入機制,也不需要註冊
- 位置資訊
- 通訊錄
- 用於追蹤或廣告的裝置識別碼
1.3 麥克風與語音辨識
「說說看」與「發音跟讀」需要使用麥克風聽取你的發音。你第一次使用時, iOS 會跳出授權詢問,你可以拒絕,只是這兩個功能會無法使用。
判斷發音是否正確時,本 App 使用 iOS 內建的語音辨識。 我們會優先使用「裝置端辨識」,讓語音不離開你的手機; 但若你的裝置或系統版本不支援裝置端辨識,iOS 會改用 Apple 的伺服器進行辨識。 這種情況下,該段語音會依照 Apple 的隱私權政策處理,我們無法取得那份語音。
1.4 跟讀錄音
使用「發音跟讀」時,你的錄音會上傳到我們自己的伺服器進行評分與逐句分析。 這是這個功能運作的必要條件——評分需要比對你的發音與示範音。
- 錄音上傳成功後,裝置上的暫存檔會立即刪除
- 伺服器上的錄音保留期限:免費方案 30 天、Pro 方案 365 天,到期後由每日排程自動刪除
- 錄音只用於產生你自己的評分與分析結果,不會用於訓練模型,也不會提供給第三方
- 伺服器由我們自行維運,不是交給第三方保管
1.5 儲存到相簿
你可以把跟讀挑戰的成績卡存到相簿。本 App 只要求「僅新增」權限, 無法讀取你相簿裡既有的任何照片。
1.6 匿名使用分析
本 App 使用 TelemetryDeck (設於德國、符合 GDPR 的分析服務)收集匿名使用統計,用來了解哪些功能真的有人用。
會收集:開啟了哪些功能、使用模式(例如測驗完成率)、App 版本與 iOS 版本、訂閱狀態(免費或 Pro)。
不會收集:任何可識別個人身分的資訊、IP 位址、使用者名稱或電子郵件、精確位置。
TelemetryDeck 使用雙重雜湊匿名化,無法從分析資料反推個別使用者。 因為沒有跨 App 追蹤行為,所以不需要顯示 App 追蹤透明度(ATT)授權視窗。
2. 我們怎麼使用這些資料
- 記錄學習進度與成就
- 產生你的發音評分與分析
- 提供學習統計
- 了解哪些功能需要改進
我們不會販售你的資料,也不會把資料用於廣告。
3. 資料儲存與安全
3.1 本機儲存
第 1.1 節列出的學習資料存在裝置的 UserDefaults,不會上傳。 也因為只存在本機,換裝置或重新安裝時,學習紀錄無法復原、也無法跨裝置同步。
3.2 會離開裝置的部分
為了讓部分功能運作,以下資料會傳送到外部服務。這是完整清單:
- 跟讀錄音 → 我們自己的伺服器(見 1.4)
- 你貼入的文字(歌詞、自訂單字)→ Google Gemini 進行分析(見 4.2)
- 發音音檔請求 → 我們的音檔服務(見 4.1)
- 訂閱狀態 → Apple 與 RevenueCat(見 4.3)
- 匿名使用統計 → TelemetryDeck(見 1.6)
- 你主動填寫的問卷 → 我們自己的伺服器(見 4.5)
3.3 刪除資料
- 本機資料:在設定中重設進度,或直接刪除 App
- 伺服器上的跟讀錄音:依 1.4 的期限自動刪除;想提前刪除請來信
4. 第三方服務
4.1 發音音檔
部分發音音檔隨 App 一起打包,另有部分是使用時即時生成的—— 例如你自己加入的單字或歌詞,事先無法預先產生。 即時生成時,會把要唸的文字送到我們的音檔服務(內部使用 ElevenLabs 語音合成)。 送出的只有要唸的文字,不含任何個人資訊。
4.2 AI 文字分析(Google Gemini)
以下功能會把你輸入或貼上的文字送到 Google 的 Gemini 服務進行分析:
- 歌詞跟讀本:解析你貼入的歌詞、標註單字
- 自訂單字:翻譯與產生例句
- 文字卡分析
這些文字會離開你的裝置、由 Google 處理,處理方式適用 Google 的隱私權政策。
我們送出的只有你提供的文字內容本身,不含你的身分資訊。
4.3 訂閱
訂閱交易由 Apple App Store 處理,付款資訊由 Apple 管理,我們無法取得你的付款細節。 我們使用 RevenueCat 管理訂閱狀態(判斷你目前是免費還是 Pro),它處理的是匿名的訂閱識別碼,不是你的付款資料。
4.4 使用分析
見 1.6 的 TelemetryDeck 說明。
4.5 問卷
App 內若邀請你填寫問卷,你填寫的內容會送到我們自己維運的伺服器。 問卷是選填的,內容用於改進 App。若你在問卷中主動留下聯絡方式,我們只會用來回覆你。
4.6 沒有廣告
本 App 沒有廣告,也沒有整合任何廣告 SDK(包含 Google AdMob)。 我們不會為了廣告而收集或分享你的資料。
5. 兒童隱私
本 App 不會刻意收集 13 歲以下兒童的個人資訊。由於本 App 不需要註冊、不收集可識別個人身分的資訊, 適合各年齡層使用。若你是家長,對 1.3 的麥克風或 1.4 的錄音上傳有疑慮,可以不使用這兩個功能, 其餘功能不受影響。
6. 你的權利
你可以隨時:
- 在 iOS 設定中撤回麥克風或語音辨識授權
- 在 App 設定中重設學習進度
- 刪除 App,移除所有本機資料
- 來信要求刪除伺服器上與你相關的跟讀錄音
7. 政策變更
本政策若有變更,會更新本頁最上方的日期。重大變更會在 App 內另行說明。
8. 聯絡我們
有任何問題請來信 wupeiwen2001@gmail.com。
Privacy Policy
Last updated: August 27, 2026
This policy explains what data Capybara Korean (the "App") processes, why, and where it goes. We have tried to describe what actually happens rather than write something vague but safe.
1. What we process
1.1 Stored only on your device
- Learning progress: letters and vocabulary learned, familiarity levels
- Quiz records: attempts, correct answers, accuracy
- Streaks: current streak, longest streak, last study date
- Achievements: unlocked badges
- Settings: sound effects, daily goal, proficiency level
1.2 What we do not collect
- Name, email, or other personally identifiable information
- Account credentials — the App has no login and requires no registration
- Location data
- Contacts
- Device identifiers used for tracking or advertising
1.3 Microphone and speech recognition
"Speak" and "Shadowing" need the microphone to hear your pronunciation. iOS will ask for permission the first time; you may decline, in which case those two features will not work.
To judge pronunciation, the App uses iOS speech recognition. We request on-device recognition first, so that audio stays on your phone. If your device or OS version does not support on-device recognition, iOS falls back to Apple's servers. In that case the audio is handled under Apple's privacy policy and we have no access to it.
1.4 Shadowing recordings
When you use Shadowing, your recording is uploaded to our own server for scoring and per-sentence analysis. This is required for the feature to work.
- The temporary file on your device is deleted immediately after a successful upload
- Retention on our server: 30 days on the free plan, 365 days on Pro, then deleted automatically by a daily job
- Recordings are used only to produce your own score and analysis. They are not used to train models and are not shared with third parties
- The server is operated by us, not entrusted to a third party
1.5 Saving to your photo library
You can save a Shadowing score card to your photos. The App requests add-only access and cannot read any existing photos.
1.6 Anonymous usage analytics
The App uses TelemetryDeck, a GDPR-compliant analytics service based in Germany, to collect anonymous usage statistics.
Collected: which features are opened, usage patterns (e.g. quiz completion), App and iOS version, subscription status (free or Pro).
Not collected: personally identifiable information, IP addresses, usernames or emails, precise location.
TelemetryDeck uses double-hashing anonymisation, so individual users cannot be identified. No App Tracking Transparency prompt is required because no cross-app tracking occurs.
2. How we use this data
- To record learning progress and achievements
- To produce your pronunciation scores and analysis
- To provide learning statistics
- To understand which features need improvement
We do not sell your data and do not use it for advertising.
3. Storage and security
3.1 Local storage
The learning data in section 1.1 is stored in iOS UserDefaults and is not uploaded. Because it is local only, your records cannot be synced across devices and cannot be recovered after reinstalling.
3.2 What does leave your device
For certain features, the following is sent to external services. This is the complete list:
- Shadowing recordings → our own server (see 1.4)
- Text you paste in (lyrics, custom words) → Google Gemini for analysis (see 4.2)
- Pronunciation audio requests → our audio service (see 4.1)
- Subscription status → Apple and RevenueCat (see 4.3)
- Anonymous usage statistics → TelemetryDeck (see 1.6)
- Surveys you choose to submit → our own server (see 4.5)
3.3 Deleting your data
- Local data: reset progress in Settings, or delete the App
- Shadowing recordings on our server: deleted automatically per section 1.4; email us to delete them sooner
4. Third-party services
4.1 Pronunciation audio
Some audio files ship with the App. Others are generated on demand — for example for words or lyrics you add yourself, which cannot be pre-generated. In those cases the text to be spoken is sent to our audio service (which uses ElevenLabs speech synthesis internally). Only the text is sent; no personal information is included.
4.2 AI text analysis (Google Gemini)
These features send text you type or paste to Google's Gemini service:
- Lyrics book: parsing lyrics you paste and annotating vocabulary
- Custom words: translation and example sentences
- Text card analysis
This text leaves your device and is processed by Google under Google's Privacy Policy.
We send only the text you provide, without identity information.
4.3 Subscriptions
Transactions are processed by the Apple App Store. Payment details are managed by Apple and we have no access to them. We use RevenueCat to manage subscription state (whether you are free or Pro); it handles anonymous subscription identifiers, not your payment data.
4.4 Analytics
See section 1.6.
4.5 Surveys
If the App invites you to complete a survey, your responses are sent to a server we operate. Surveys are optional. If you choose to leave contact details, we use them only to reply to you.
4.6 No advertising
The App contains no advertising and integrates no advertising SDK (including Google AdMob). We do not collect or share your data for advertising purposes.
5. Children's privacy
The App does not knowingly collect personal information from children under 13. It requires no registration and collects no personally identifiable information. If you are a parent concerned about the microphone (1.3) or recording uploads (1.4), those two features can simply be left unused; everything else works without them.
6. Your choices
- Revoke microphone or speech recognition permission in iOS Settings at any time
- Reset your learning progress in the App's settings
- Delete the App to remove all local data
- Email us to request deletion of your Shadowing recordings from our server
7. Changes to this policy
If this policy changes, the date at the top of this page will be updated. Significant changes will also be noted in the App.
8. Contact
Questions? Email wupeiwen2001@gmail.com.