WordBites — 隐私政策 / Privacy Policy
生效日期 / Effective date: 2026-07-11
中文English
中文
WordBites（“本应用”）是一款英语单词学习应用。我们重视你的隐私。本政策说明我们收集哪些信息、如何使用、与谁共享，以及你拥有的权利。
1. 我们收集的信息
- 账号信息：你可选择以邮箱或 Google 账号登录；此时我们会处理你的邮箱、显示名。你也可以匿名使用，此时仅生成一个随机匿名标识。
- 学习数据：你的单词卡片、生词本、学习进度、统计、金币、宠物状态等。这些数据保存在本地，并在你登录后同步到云端（Google Firebase），以便多设备同步。
- 音频：当你使用「实时字幕」「发音训练」「影子跟读」「口语评分」「AI 语音通话」「AI 视频对话」等功能时，应用会通过麦克风录音。音频用于转写成文字 / 发音评分 / 实时语音对话，会发送给相应的语音识别或 AI 语音服务处理（见第 3 节）。我们不会长期保存原始录音。
- 图片 / 摄像头：当你使用「拍照 / 批量识图 / Live OCR / 悬浮识词」时，应用会拍照或读取你选择的图片，用于文字识别（OCR）。图像识别在设备本地完成；识别出的文字可能被发送给 AI 服务用于生成卡片。
- 你输入的内容：你主动输入或粘贴的文字（如翻译练习、AI 对话、文章、新闻搜索词），会发送给相应服务以生成结果；新闻搜索词会连同国家 / 语言发送给新闻聚合服务（RapidAPI）以返回相关标题与摘要。
- 设备标识：为防止滥用（例如通过卸载重装反复领取免费试用或新用户额度），我们会收集你的 Android 设备标识（SSAID），仅用于防欺诈的设备指纹校验，不用于广告或跨应用追踪。
- 设备与使用信息：通过 Firebase Analytics 收集匿名的使用统计与崩溃信息，用于改进应用。
2. 我们如何使用这些信息
用于：提供和同步学习功能、生成 AI 内容（卡片 / 释义 / 对话 / 评分）、语音转文字、图片文字识别、改进应用稳定性与体验。我们不会出售你的个人信息。
3. 第三方共享
为实现上述功能，部分数据会发送给以下服务商处理：
- Google Firebase（身份认证、云数据库、分析、云函数）—— 账号与学习数据。
- 通过我们的服务器代理转发给：OpenAI（AI 文本生成；以及「AI 语音通话」的实时音频语音处理）、DeepSeek、Google Gemini（AI 文本生成）、Deepgram（语音转文字）、RapidAPI（新闻标题与摘要）、Tavus（AI 视频对话）等。发送的内容仅限实现该功能所必需的数据（如你的录音、你输入的文字）。
- Google Play 结算（Pro / Max 订阅与内购）—— 订阅与内购由 Google Play 处理，我们不收集、不接触你的支付卡号等支付信息。
这些服务商各自有其隐私政策，受其条款约束。跨境处理：由于上述部分服务商（如 OpenAI、Google 等）位于美国等你所在国家 / 地区之外，为实现相应功能，你的相关数据可能被传输至境外并在当地处理。
4. 权限说明
| 权限 | 用途 |
|---|
| 麦克风 | 实时字幕、发音 / 口语评分、AI 视频对话的录音 |
| 摄像头 | 拍照识词、AI 视频对话 |
| 照片 / 媒体 | 批量识图建卡时读取你选择的图片 |
| 通知 | 复习提醒、学习提示 |
| 悬浮窗 / 截图取词 | 悬浮识词气泡；截图后读取你自己的截图用于取词识别（不录屏、不读取其他应用内容） |
所有权限都在你使用对应功能时才请求；你可以随时在系统设置中撤销。
5. AI 生成内容说明
本应用的部分内容（单词释义、例句、对话、文章、评分等）由人工智能生成。AI 生成的内容可能存在错误或不准确，仅供学习参考，不构成专业建议。
5.5 用户分享内容（共享词包广场）
- 「共享广场」允许你自主选择把自己创建的词包（自建词包或 AI 定制词包）发布给其他用户。只有在你主动点击「分享到广场」并确认后，相关内容才会被上传。
- 你发布的内容包括：词包名称、你填写的作者昵称、词包简介，以及词包内的全部词卡（单词、释义、例句等）。这些内容一经审核通过即对所有用户公开可见。
- 请不要在分享内容（含昵称）中包含任何个人敏感信息。你的账号 uid 仅用于服务端审核与下架管理，不会在广场上展示。
- 所有分享内容采用先审后发：发布后需经人工审核通过才会上架。你或其他用户可举报违规内容，我们会进行核查并可下架。
- 你可随时联系我们要求下架你已发布的词包。
5.6 匿名众包词库
- 为帮助我们完善内置词库与释义质量，当你新增一张单词卡时，应用会把该英文单词及其释义文本匿名上报到一个聚合词库池。
- 上报内容仅包含英文单词与释义文本，不包含你的账号标识（uid）、个人笔记或任何可识别你身份的信息；服务端只做“该词被多少用户加过”的聚合统计。
- 该数据用于发现高频生词、改进推荐与内置词库，经人工审核后可能纳入未来版本的内置 / 精选词库。它不会与你的身份关联，也不会被出售。
- 你可以在 设置 → 隐私 中随时关闭“匿名贡献词汇”，关闭后将不再上报。
6. 数据存储、保留与删除
- 学习数据保存在你的设备本地，并（登录后）保存在 Firebase 云端。
- 你可以在 设置 → 隐私 → 删除账号 中删除你的账号及云端数据。删除后你的个人数据（学习数据、昵称、共享词包、举报记录等）将被移除且不可恢复；用于防欺诈的设备指纹会解除与你账号的关联。
- 保留例外：为遵守法律与业务要求（如退款、税务、订阅有效性校验），与购买相关的交易凭证可能在账号删除后按法定期限继续保留。
- 你也可以通过卸载应用清除本地数据。
6.5 你的权利
在适用法律（如《澳大利亚隐私法》、欧盟 GDPR 等）允许的范围内，你有权：
- 访问 / 更正：多数学习数据可在应用内直接查看与修改；如需其他协助，可通过下方邮箱联系我们。
- 删除：随时在「设置 → 隐私 → 删除账号」删除账号及数据，或访问 数据删除页面 提交请求。
- 选择退出：在「设置 → 隐私」中关闭「匿名贡献词汇」与分析 / 崩溃数据上报。
- 撤回同意 / 反对处理：停止使用相应功能即可停止相关数据的收集。
- 投诉：你有权向你所在地的数据保护 / 隐私监管机构投诉。
7. 儿童隐私
本应用不面向 13 周岁以下儿童设计，我们不会有意收集其个人信息。
8. 安全
我们采取合理措施保护你的数据：第三方 API 密钥仅存于服务器端、不随应用分发；云端访问受身份认证保护。但任何传输或存储方式都无法保证绝对安全。
9. 内容来源与署名
本应用的部分学习内容来自公有领域 / 开放授权来源，在此依其许可要求署名致谢：
- 电台：美国之音「VOA Learning English」直播流。美国之音（Voice of America）为美国联邦机构，其自制内容属公有领域。© Voice of America / voanews.com。本应用仅指向其官方公开直播流播放，不转存、不录制、不再转播。
- 经典名著（阅读 / 朗读）：公有领域文本来自 Project Gutenberg（gutenberg.org）；公有领域有声书来自 LibriVox（librivox.org）。
- 新闻：仅聚合各媒体公开 RSS 的标题与摘要并跳转原站阅读，不复制、不转存全文，不再托管其图片。
- 词典与百科：离线释义部分基于开放词典 ECDICT；百科精选来自维基百科（Wikipedia，CC BY-SA）。
若你认为本应用的某项内容侵犯了你的权利，请通过下方邮箱联系我们，我们会及时处理。
10. 政策变更
我们可能更新本政策；重大变更会在应用内或通过本页面告知。
11. 联系我们
如有隐私相关问题，请联系：wordbitescontactus@gmail.com
English
WordBites ("the App") is an English vocabulary-learning app. We respect your privacy. This policy explains what we collect, how we use it, who we share it with, and your rights.
1. Information we collect
- Account: You may sign in with email or Google (we then process your email and display name), or use the App anonymously (a random anonymous ID).
- Learning data: Your word cards, study progress, stats, coins, pet state. Stored locally and, once signed in, synced to the cloud (Google Firebase).
- Audio: Features like Live Subtitles, Pronunciation, Shadow Reading, Speech Scoring, AI Voice Call and AI Video Chat record from your microphone. Audio is sent to speech-recognition or AI voice providers for transcription / scoring / real-time voice conversation (see §3). We do not retain raw recordings long-term.
- Images / camera: OCR features capture a photo or read images you pick, for on-device text recognition. Recognized text may be sent to AI services.
- Content you enter: Text you type or paste (translations, AI chat, articles, news search terms) is sent to the relevant service to produce results; news search terms are sent, together with country / language, to a news-aggregation service (RapidAPI) to return relevant headlines and summaries.
- Device identifier: To prevent abuse (e.g. repeatedly claiming free trials or new-user credits by reinstalling), we collect your Android device identifier (SSAID), used only for anti-fraud device fingerprinting — not for advertising or cross-app tracking.
- Device & usage: Anonymous usage stats and crash data via Firebase Analytics, to improve the App.
2. How we use it
To provide and sync learning features, generate AI content, transcribe speech, recognize text in images, and improve the App. We do not sell your data.
3. Third-party sharing
To deliver these features, some data is processed by: Google Firebase (auth, database, analytics, functions); and — via our server proxy — OpenAI (AI text; and real-time audio processing for AI Voice Call), DeepSeek, Google Gemini (AI text), Deepgram (speech-to-text), RapidAPI (news headlines & summaries), Tavus (AI video); and Google Play Billing for subscriptions and in-app purchases — payments are handled by Google Play and we never collect or see your card / payment details. Only data necessary for the feature is sent. Each provider has its own privacy policy. International processing: because some of these providers (e.g. OpenAI, Google) are located outside your country (e.g. in the United States), your relevant data may be transferred abroad and processed there to deliver the feature.
4. Permissions
Microphone (recording for transcription / scoring / video chat), Camera (OCR, video chat), Photos / Media (images you pick for OCR), Notifications (review reminders), Overlay / Screenshot lookup (floating word bubble; reading your own screenshots to extract words — no screen recording, no reading other apps' content). Permissions are requested only when you use the related feature and can be revoked anytime in system settings.
5. AI-generated content
Some content (definitions, examples, dialogues, articles, scores) is generated by AI and may be inaccurate. It is for learning reference only.
5.5 User-shared content (Shared Packs plaza)
- The "Shared Plaza" lets you optionally publish word packs you created (self-built or AI-customized) to other users. Content is uploaded only after you tap "Share to plaza" and confirm.
- What you publish: the pack name, the author nickname you enter, the pack description, and all word cards inside it (words, meanings, examples). Once approved this is publicly visible to all users.
- Do not include personal or sensitive information in shared content (including the nickname). Your account uid is used server-side only for review and takedown management and is never displayed in the plaza.
- All shared content is reviewed before going live. You or other users can report content; we will review and may remove it.
- You may contact us at any time to take down a pack you published.
5.6 Anonymous crowdsourced vocabulary
- To help us improve our built-in word lists and definition quality, when you add a word card the App anonymously reports that English word and its definition text to an aggregated vocabulary pool.
- The report contains only the English word and definition text — it does not include your account identifier (uid), personal notes, or anything that identifies you. The server only aggregates how many users added each word.
- This data is used to surface common words and improve recommendations and built-in lists, and after human review may be included in future built-in / curated lists. It is never linked to your identity and is never sold.
- You can turn off "Contribute vocabulary anonymously" anytime in Settings → Privacy.
6. Storage, retention & deletion
Learning data is stored on your device and (when signed in) in Firebase. You can delete your account and cloud data in Settings → Privacy → Delete account; your personal data (learning data, nickname, shared packs, reports, etc.) is permanently removed and the anti-fraud device fingerprint is unlinked from your account. Retention exception: transaction records related to purchases may be retained after account deletion for the period required by law and business needs (e.g. refunds, tax, subscription validation). Uninstalling the App clears local data.
6.5 Your rights
To the extent applicable law allows (e.g. the Australian Privacy Act, EU GDPR):
- Access / correct: most learning data can be viewed and edited directly in-app; email us at the address below for other assistance.
- Delete: delete your account and data anytime via Settings → Privacy → Delete account, or request deletion at the data-deletion page.
- Opt out: turn off "Contribute vocabulary anonymously" and analytics / crash reporting in Settings → Privacy.
- Withdraw consent / object: stop using a feature to stop the related data collection.
- Complain: you may lodge a complaint with your local data-protection / privacy authority.
7. Children
The App is not directed to children under 13; we do not knowingly collect their data.
8. Security
Third-party API keys are kept server-side only and not shipped in the App; cloud access is protected by authentication. No method is 100% secure.
9. Content sources & attribution
Some learning content comes from public-domain / openly-licensed sources, credited here per their license terms:
- Radio: the "VOA Learning English" live stream. Voice of America is a U.S. federal agency and its original content is in the public domain. © Voice of America / voanews.com. The App only points to VOA's official public live stream for playback — it does not store, record, or rebroadcast it.
- Classic literature (reading / audio): public-domain texts from Project Gutenberg (gutenberg.org); public-domain audiobooks from LibriVox (librivox.org).
- News: only public RSS titles and snippets are aggregated with a link out to the original site; full article text is not copied or stored, and publisher images are not re-hosted.
- Dictionary & encyclopedia: offline definitions are based on the open ECDICT dictionary; featured encyclopedia entries come from Wikipedia (CC BY-SA).
If you believe any content in the App infringes your rights, contact us at the email below and we will address it promptly.
10. Changes
We may update this policy; material changes will be notified in-app or here.
11. Contact
Privacy questions: wordbitescontactus@gmail.com