隐私政策Privacy Policy
最后更新：2026 年 7 月 12 日。Echo 是一款 macOS 语音输入工具，默认优先在本机完成语音识别。 Last updated: July 12, 2026. Echo is a macOS voice input tool that prioritizes local speech recognition by default.
1. 我们收集或处理的数据
为了提供语音输入功能，Echo 可能在你的设备上处理以下数据：
- 麦克风音频：用于把你说的话转成文字。
- 转写文本：用于显示、注入当前输入框、可选保存历史，以及进行本地后处理。
- AI 指令内容：当你按下 AI 指令键时，可能包括本次转写文本、选中文本、你明确要求处理的剪贴板文本、搜索摘要或屏幕可见文字。
- 个人术语词典和设置：例如快捷键、麦克风、模型路径、语言和 AI 补强偏好。
- 诊断信息：例如模型是否就绪、权限状态、错误类型和权益同步状态。诊断信息用于排障、同步订阅状态和改进稳定性，不代表默认上传你的录音或完整转写历史。
2. 本地优先处理
Echo 默认优先使用本地语音识别模型。能在本机完成的识别、后处理和设置保存，原则上不需要发送到我们的服务器。
3. 可选云端与 Pro 托管能力
如果你主动开启云端补强、AI 指令或 Pro 托管能力，Echo 可能会处理转写文本、选中文本、必要上下文、搜索摘要、剪贴板文本或屏幕可见文字。自配 API Key 时，相关文本会发送给你配置的云端服务商，例如 DashScope / Qwen、OpenAI 兼容服务或其他你选择的服务。使用 Pro 托管能力时，相关请求会经 Echo 中转服务处理，用于完成请求、计量额度和防滥用。云端能力用于改善低置信度结果、搜索后回答、改写文本、翻译、总结、工作套件或理解当前屏幕。你可以在设置中关闭相关功能。
4. 系统权限
Echo 会请求 macOS 权限以完成核心功能：
- 麦克风权限用于录音。
- 辅助功能权限用于把文字输入到当前输入框。
- 输入监控权限用于监听全局快捷键，尤其是按住说话模式。
- 屏幕录制权限仅用于你明确要求“AI 看屏幕”时读取当前窗口可见文字。
5. 数据保存与删除
设置、术语词典、本地模型状态和可选转写历史通常保存在你的 Mac 上。你可以在 App 内修改设置、清理历史，或删除本地应用数据。卸载 App 不一定会自动删除所有 Application Support 数据，你也可以手动清理。
6. 第三方服务
当你使用可选云端补强、AI 搜索问答、订阅购买或 App Store 分发能力时，相关第三方可能会按其隐私政策处理必要数据。包括但不限于 Apple App Store、你配置的 AI 服务商，以及你配置的搜索服务商。请同时阅读这些第三方的隐私政策。
7. 儿童隐私
Echo 不面向儿童设计。若你认为儿童向我们提供了个人信息，请联系我们处理。
8. 政策变更
我们可能会根据产品功能、法律要求或 App Store 审核要求更新本政策。重大变更会在网站或 App 内提示。
9. 联系我们
如对隐私政策有问题，请联系：support@flowlab.im。
1. Data We Collect or Process
To provide voice input, Echo may process the following data on your device:
- Microphone audio, used to convert your speech into text.
- Transcript text, used for display, insertion into the active field, optional history, and local cleanup.
- AI command content, which may include the current transcript, selected text, clipboard text you explicitly ask Echo to process, search summaries, or visible screen text.
- Personal dictionary and settings, such as shortcuts, microphone choice, model path, language, and AI correction preferences.
- Diagnostic information, such as model readiness, permission state, error type, and entitlement sync status. Diagnostics help with troubleshooting, subscription sync, and reliability; they do not mean your audio or full transcript history is uploaded by default.
2. Local-First Processing
Echo prioritizes local speech recognition by default. Recognition, cleanup, and settings that can be handled on your Mac generally do not need to be sent to our servers.
3. Optional Cloud Correction
If you enable cloud correction, AI commands, or Pro hosted features, Echo may process transcript text, selected text, required context, search summaries, clipboard text, or visible screen text. With your own API key, text is sent to the cloud provider you configure, such as DashScope / Qwen, an OpenAI-compatible service, or another service you choose. When you use Pro hosted features, related requests pass through Echo services to complete the request, meter credits, and prevent abuse. Cloud features are used for low-confidence cleanup, search-backed answers, rewriting, translation, summarization, work suites, or screen understanding. You can turn related features off in settings.
4. System Permissions
Echo requests macOS permissions to provide core features:
- Microphone permission is used to record audio.
- Accessibility permission is used to insert text into the active field.
- Input Monitoring permission is used to listen for global shortcuts, especially hold-to-talk.
- Screen Recording permission is used only when you explicitly ask Echo to read visible text from the current window.
5. Storage and Deletion
Settings, dictionary entries, local model state, and optional transcript history are generally stored on your Mac. You can change settings, clear history, or delete local app data. Uninstalling the app may not automatically remove all Application Support data.
6. Third-Party Services
When you use optional cloud correction, search-backed answers, subscriptions, purchases, or App Store distribution, relevant third parties may process necessary data under their own privacy policies, including Apple App Store, the AI provider you configure, and the search provider you configure.
7. Children’s Privacy
Echo is not designed for children. If you believe a child has provided personal information to us, please contact us.
8. Changes
We may update this policy as the product, legal requirements, or App Store review requirements change. Material changes may be announced on the website or in the app.
9. Contact
Questions about this policy can be sent to support@flowlab.im.