演讲镜子 (Speech Mirror) 隐私政策

更新日期：2026年07月14日

「演讲镜子」App（以下简称“本应用”）极其重视您的隐私。本政策旨在说明我们如何处理您的个人信息。本应用作为一款100% 本地运行（On-Device）的演讲与口播自训工具，致力于提供完全私密的离线排练诊断环境。

第一部分：信息收集与使用声明

本应用不需要您注册任何账户，也无需登录。我们不会收集您的姓名、电话号码、电子邮件地址等任何个人身份识别信息。

第二部分：相机与麦克风权限

当您使用“开始排练”功能时，应用会向您申请授权访问相机与麦克风：

• 麦克风权限：仅用于捕捉您的演讲发言语音，并在本地利用 Apple Speech 框架进行离线文本转写、语速（WPM）及语气词分析。
• 相机权限：仅用于获取视频流，并在本地进行视线对齐（眼神注视度）及低头姿态分析。

声明：本应用绝不会向云端上传 any 音频、视频、转写文本或分析报告。所有录制的排练视频文件均以绝对离线方式存储在您设备本地的安全沙盒内。

第三部分：人脸及视线数据处理说明（100%本地与即时销毁）

本应用在分析眼神交流和排练仪态时会涉及面部几何特征点的处理，我们对此郑重作出以下合规合法的政策约束：

1. 处理的数据类型：本应用仅在本地使用苹果原生 Vision 框架检测您眼部、头部的几何坐标点（视线偏离值、头部倾斜度）。我们绝不收集、不扫描、也不读取任何面部 3D 深度图、面部生物特征等可用于识别您个人身份的面部数据。

2. 处理目的：上述面部几何坐标的提取仅用于在排练中计算“眼神专注度”和“低头频率”，并在排练结束后在诊断报告中向您展示得分与改进对策。

3. 共享与披露（第三方政策）：本应用绝不会将面部或视线数据与任何第三方共享。没有第三方 SDK，也无任何数据同步后台。

4. 数据存储与保留周期：面部几何特征坐标在本地内存（RAM）中进行即时、临时的帧计算。保留周期为 0 秒，每帧计算结束后立即在内存中进行销毁，不写入本地磁盘文件，也不上传到云端服务器。

第四部分：第三方服务

本应用完全离线运行，不包含任何第三方分析工具或广告系统 SDK。没有任何第三方能够获取您的任何行为数据或设备信息。

第五部分：联系我们

如果您对本隐私政策有任何疑问，请通过应用市场开发者通道与我们取得联系。

Speech Mirror Privacy Policy

Last Updated: July 14, 2026

"Speech Mirror" (hereinafter referred to as "the App") respects your privacy. This privacy policy explains our practices regarding user data. Built as a 100% on-device (offline) rehearsal and coaching utility, the App provides a fully sandboxed, local environment to safeguard your pitch and speech data.

1. Data Collection and Usage

The App does not require any user registration or account login. We do not collect, request, or store any personally identifiable information (PII) such as your name, email, or telephone number.

2. Camera and Microphone Access

To deliver rehearsal analytics, the App requests permissions to access your device's Camera and Microphone:

• Microphone Permission: Used solely to capture your voice locally for offline speech-to-text transcription, pacing (WPM), and filler word analysis via Apple's built-in local Speech Framework.
• Camera Permission: Used solely to capture the live video feed to analyze gaze focus (eye contact) and head tilt (looking down posture) locally.

Note: Rehearsal video files are stored strictly inside the App's local container directory on your device. The App never uploads any audio, video, text transcripts, or reports to external cloud servers.

3. Face Data Processing (100% Local & Transient)

When analyzing eye contact and rehearsal posture, the App processes geometric landmarks of your face. We enforce strict compliance policies regarding this data:

A. Types of Data Processed: The App utilizes Apple’s local Vision framework to compute standard eye coordinate vectors and head tilt angles from the camera feed. We do not collect, scan, or access any 3D facial depth mapping or face biometric signatures that can identify a specific user.

B. Purpose of Processing: The facial geometry vectors are computed solely to evaluate gaze focus and head tilt posture metrics during your presentation runs.

C. Sharing & Disclosure (Third-Party Policy): We do not share, sell, or disclose your facial/gaze coordinates to any third parties. No analytics trackers or ad SDKs are bundled with the App.

D. Storage & Retention Period: All facial calculations are processed transiently in the active RAM frame-by-frame. The retention period is 0 seconds. Data is immediately discarded from memory after the current frame is processed, and is never written to disk or transmitted to any server.

4. Third-Party Services

The App runs entirely offline. It does not integrate any third-party analytics trackers, cloud-based AI endpoints, or advertisement networks.

5. Contact Us

If you have any questions regarding this Privacy Policy, please contact the developer via the App Store feedback channel.