您的音频与文字内容从不离开您的设备。语音识别与翻译全部在您的 Mac 上本地完成，TransLive 不收集、不上传任何数据。
什么都不收集。TransLive 没有账号系统、没有分析统计、没有广告、没有崩溃上报，也不包含任何分析、广告或跟踪类第三方 SDK。应用内置的第三方组件（MLX、llama.cpp、ONNX Runtime 等）是用于本地推理的开源库；正常使用时推理在设备上完成。
在您授权后，TransLive 实时采集麦克风或系统音频用于本地语音识别。音频数据仅在内存中流式处理，处理后即丢弃——不写入磁盘、不上传到任何服务器。
为了「历史记录」功能，识别与翻译的文字结果会保存在您本机的应用数据目录中——仅存于本机、仅您可见。您可以随时在应用内搜索、删除或全部清空。导出的笔记 / 字幕文件保存到您自己选择的位置。
正式分发版的语音识别与翻译模型（Apache-2.0 许可）已内置，正常使用无需联网；音频、转写和翻译内容不会上传。应用的 Mac App Store 沙盒权限未申请网络访问。
无。
如本政策发生变更，我们会在本页更新并注明新的生效日期。
如有任何疑问，请联系：bbzlkcn@gmail.com
Your audio and text never leave your device. Speech recognition and translation run entirely on your Mac — TransLive collects and uploads nothing.
Nothing. TransLive has no accounts, no analytics, no ads, no crash reporting, and no analytics/advertising/tracking SDKs of any kind. The third-party components it bundles (MLX, llama.cpp, ONNX Runtime, etc.) are open-source inference libraries used on-device during normal operation.
With your permission, TransLive captures microphone or system audio for on-device speech recognition. Audio is processed as an in-memory stream and discarded immediately — it is never written to disk or uploaded to any server.
To power the History feature, recognition and translation text is stored in the app's local data folder on your Mac — local only, visible only to you. You can search, delete, or clear them at any time inside the app. Exported notes and subtitle files are saved to locations you choose.
The Mac App Store build bundles its Apache-2.0-licensed speech and translation models, so normal use requires no network connection. Audio, transcripts, and translations are never uploaded, and the App Store sandbox does not request network access.
None.
If this policy changes, we will update this page with a new effective date.
Questions? Email bbzlkcn@gmail.com