能。在「录音」页用 iPhone 录，锁屏也会继续；也可以导入已有的录音。

叫「回声」，需要 watchOS 11。没自动装上的话，在 iPhone 的 Watch App 里安装。

长按表盘 › 编辑 › 复杂功能，选「回声」。

在「设置 › 示例」里加载示例数据，看完可以一键清除。

手表上有红点和计时，iPhone 顶部有红点，系统也会显示麦克风提示。只有你点了开始才会录。

很久没人说话（默认 45 分钟）、手表电量低于 10%、iPhone 电量低于 5% 时，会自动停止并保存。

会，这是系统规定。打断前的内容都会保存；接不上时手表会提醒你。

Series 9 实测熄屏连录约 13 小时，用电约 74%，因机型和用法而异。

每 10 分钟一段，在 iPhone 旁边就自动传。iPhone 收到后，手表才删掉这一段。

每小时约 14 MB。旧音频会按设置自动清理，文字一直保留。

中文（普通话、台湾）、粤语、英语、日语、韩语、法语、德语、西班牙语、意大利语、葡萄牙语。

在「设置 › 人物与声纹」里读 20 秒文字录声纹。录别人的声纹前，请先征得对方同意。

长按句子可以编辑，点说话人标签可以改人。要紧的地方请回听原声。

默认每晚 21:00 后自动整理，每周一出周报；也可以随时点「整理」。

都在本机完成。支持 Apple 智能的 iPhone 用 Apple 智能，内存 6 GB 以上可以下载本地模型；其他情况由设备上的 AI 从原话里挑出重点，不生成新内容。App Store 地区为中国大陆时，只用挑重点的方式。

从 Hugging Face 下载开源模型 Qwen3，你确认后才开始，只下载、不上传。

一次买断，没有内购和订阅。

不用。更新到最新版，全部功能都能用。

到 reportaproblem.apple.com 申请。

不会发给开发者或任何第三方，转写和整理都在你的设备上。文字和声纹会随你自己的 iCloud 或电脑备份保存，音频默认不进备份。详见隐私政策。

「设置 › 存储」里导出全部文字；「设置」里可以删除全部数据。

在「设置 › 隐私」里打开面容 ID 锁定。

Yes. Record on iPhone from the Recordings tab, even with the screen locked, or import recordings you already have.

It's called WristEcho and needs watchOS 11. If it didn't install automatically, install it from the Watch app on your iPhone.

Touch and hold the watch face › Edit › Complications, then choose WristEcho.

Load sample data in Settings › Sample, and clear it with one tap when you're done.

The watch shows a red dot and timer, the iPhone shows a red dot at the top, and the system shows its mic indicator. It only records after you tap start.

After a long silence (45 minutes by default), when the watch battery drops below 10%, or when the iPhone drops below 5%, it stops and saves.

Yes, the system requires it. Everything recorded before is saved, and the watch reminds you if it can't resume.

In our overnight test, Apple Watch Series 9 recorded about 13 hours with the screen off and used about 74% of its battery. It varies by model and use.

In 10-minute segments, sent automatically when the iPhone is nearby. The watch deletes a segment only after the iPhone has it.

About 14 MB per hour. Old audio is cleaned up based on your settings; text is always kept.

Chinese (Mandarin, Taiwan), Cantonese, English, Japanese, Korean, French, German, Spanish, Italian and Portuguese.

Read a short text for 20 seconds in Settings › People & Voiceprints. Ask other people for consent before enrolling their voice.

Touch and hold a line to edit it, or tap the speaker label to change who said it. Check important details against the original audio.

By default it's written after 9 PM each evening, with a weekly report every Monday. You can also tap Summarize at any time.

All on your device. iPhones with Apple Intelligence use it, and iPhones with 6 GB of memory or more can download a local model; otherwise on-device AI picks the key sentences you said, without generating new text. When your App Store region is China mainland, only the key-sentence method is used.

The open-source Qwen3 model is downloaded from Hugging Face only after you confirm. Nothing is uploaded.

It's a one-time purchase, with no in-app purchases or subscription.

No. Update to the latest version and every feature is available.

Go to reportaproblem.apple.com.

They are never sent to us or any third party; transcription and reports happen on your devices. Transcripts and voiceprints are included in your own iCloud or computer backup, and audio is left out of backups by default. See the Privacy Policy.

Export all text in Settings › Storage, and delete all data in Settings.

Turn on Lock with Face ID in Settings › Privacy.