TripOral（随口译）隐私政策
生效日期：2026 年 7 月 6 日
一句话概括：TripOral 不需要账号，App 本身不收集、不上传你的个人数据。使用本地引擎时，语音和文字不离开你的设备；只有当你主动配置外部模型时，待翻译的文本才会发送给你自己选择的服务商。
1. 我们是谁
TripOral（随口译）是一款 iOS 实时语音翻译应用。本政策说明 App 处理哪些数据、在哪里处理、以及你的控制方式。
2. 我们不收集什么
- 无需注册：App 没有账号系统，不收集姓名、邮箱、电话等身份信息。
- 无追踪：App 不包含第三方广告 SDK 或第三方分析/统计 SDK，不做跨应用追踪。
- 无自建服务器上传：我们没有接收用户内容的服务器；你的翻译内容不会发送给我们。
3. 麦克风与语音识别
翻译需要使用麦克风采集你的语音，并通过 Apple 的语音识别（Speech 框架）转成文字。App 会优先使用设备端识别；当设备端识别模型不可用时，会自动改用系统听写通道，此时音频可能被发送至 Apple 的服务器处理，受 Apple 隐私政策约束。麦克风仅在你主动开始翻译/对话时启用。
4. 翻译引擎与数据去向
- 系统本地引擎（默认）：使用 iOS 系统离线翻译与系统语音合成。下载离线语言包后，识别、翻译、朗读均在设备上完成，内容不离开设备。
- 本地增强（Gemma 端侧模型）：模型文件下载到设备后，推理完全在设备上进行，内容不离开设备。模型文件经 Apple 的分发通道（Background Assets）或你自行配置的地址下载。
- 外部模型（可选，需你主动配置）：当你填入自己的 API 服务商（如 OpenAI 兼容接口）并启用时，待翻译的文本（不含音频）会发送到你配置的服务商，受该服务商的隐私政策约束。API 密钥仅保存在设备钥匙串中，绝不上传。
5. 历史记录
翻译与会话记录仅保存在你的设备本地。你可以随时删除单条或全部记录，也可以开启无痕模式，不保存任何记录。删除 App 即删除全部本地数据。
6. 儿童隐私
App 面向一般用户，不针对 13 岁以下儿童设计，也不有意收集儿童的个人信息（如上所述，App 不收集任何人的个人信息）。
7. 政策变更
若本政策发生实质变更，我们会在本页面更新并调整生效日期。重大变更会在 App 内提示。
8. 联系我们
对本政策有任何疑问，请联系：catregonggl@gmail.com
TripOral Privacy Policy
Effective date: July 6, 2026
In one sentence: TripOral requires no account and the app itself does not collect or upload your personal data. With the local engine, audio and text never leave your device; only if you deliberately configure an external model is the text to be translated sent to the provider you chose.
1. Who we are
TripOral is an iOS real-time voice translation app. This policy explains what data the app processes, where it is processed, and what you control.
2. What we do not collect
- No sign-up: the app has no account system and collects no name, email, phone number or other identity data.
- No tracking: the app contains no third-party advertising or analytics SDKs and does no cross-app tracking.
- No upload to our servers: we operate no servers that receive user content; your translations are never sent to us.
3. Microphone and speech recognition
Translation uses the microphone to capture your speech and Apple's Speech framework to transcribe it. The app prefers on-device recognition; when the on-device model is unavailable it automatically falls back to the system dictation channel, in which case audio may be processed on Apple's servers under the Apple Privacy Policy. The microphone is only active while you are actively translating or in a conversation session.
4. Translation engines and where data goes
- System local engine (default): uses iOS offline translation and system text-to-speech. With offline language packs downloaded, recognition, translation and speech all run on the device; content never leaves it.
- Local boost (on-device Gemma model): after the model files are downloaded to the device, inference runs entirely on-device; content never leaves it. Model files are downloaded via Apple's distribution channel (Background Assets) or an address you configure yourself.
- External model (optional, configured by you): if you enter your own API provider (e.g. an OpenAI-compatible endpoint) and enable it, the text to be translated (never audio) is sent to that provider and is subject to its privacy policy. Your API key is stored only in the device keychain and is never uploaded.
5. History
Translation and conversation history is stored only locally on your device. You can delete individual items or everything at any time, or use incognito mode to keep no history at all. Deleting the app deletes all local data.
6. Children's privacy
The app is intended for a general audience, is not designed for children under 13, and does not knowingly collect personal information from children (as described above, it collects personal information from no one).
7. Changes to this policy
If this policy changes materially, we will update this page and the effective date. Significant changes will be announced in the app.
8. Contact
Questions about this policy: catregonggl@gmail.com