同声 · Simul
把你听到的话实时变成双语字幕:原文在上,译文在下。识别和翻译全程在这台设备上完成 —— 飞行模式下照常工作。
第一次使用需要先下载语言模型。打开 App 会有引导页带你装;如果你跳过了,去「设置 → 添加语言」也能装。这一步需要联网,装好之后就再也不需要了。
取决于语言和你的网络。系统不提供总大小的预估,所以我们不在界面上编一个数字给你看。模型由 iOS 系统统一管理、多个 App 共享,不计入「同声」的安装体积。
能识别的语言由你的 iOS 版本决定,不同系统版本支持的范围不一样。把系统升到最新通常能多出几种。
是的,这是 iOS 的限制,不是我们的。超出时系统会释放最久没用的那个。你需要时再装回来即可。
在 iOS 上,App 无法直接采集其他 App 的声音,通话内容更是永远拿不到 —— 这是系统层面的限制。现在的做法是:把视频外放,让「同声」通过麦克风听。
那一行表示「这句话还没说完,内容可能还会变」。说完之后它会定稿变成白色。这样你一眼就知道哪句可以信、哪句还在变。
不能。watchOS 系统本身不提供语音识别和翻译能力,所以手表版只负责显示 iPhone 推来的字幕,以及远程开始 / 停止。iPhone 必须在附近并且开着。
都不会。音频实时处理完就丢掉,字幕只在内存里,退出这一屏就没有了。详见隐私政策。
遇到问题、想要某个语言、或者只是想说点什么,请发邮件到 dengzhaofun@gmail.com。
报告字幕相关的问题时,如果能做这一步会很有帮助:打开「设置」,连续点击「版本」那一行 7 次,打开出现的「显示诊断信息」,然后把字幕页底部那一行截图发给我们。
Simul turns what you hear into bilingual captions in real time: the original on top, the translation below. Recognition and translation run entirely on this device — it works in Airplane Mode.
Language models have to be downloaded first. The app walks you through it on first launch; if you skipped that, go to Settings → Add a language. This step needs a network connection; after that, you never need one again.
It depends on the language and your connection. The system does not report a total size in advance, so we do not invent a number to show you. Models are managed by iOS and shared across apps, so they do not count toward Simul's install size.
Which languages can be recognised is decided by your version of iOS, and it differs between releases. Updating to the latest system version usually adds a few.
Yes — that is an iOS limit, not ours. Beyond that the system releases the least recently used one. Install it again whenever you need it.
On iOS an app cannot capture another app's audio, and phone call audio is never available — both are system-level restrictions. For now: play the video out loud and let Simul listen through the microphone.
That line means "this sentence isn't finished; the text may still change." Once the sentence settles it turns white. So you always know which lines you can trust.
No. watchOS itself provides no speech recognition or translation, so the watch app only displays captions pushed from your iPhone and lets you start and stop remotely. Your iPhone must be nearby and running.
Neither. Audio is processed live and discarded; captions live only in memory and are gone when you leave the screen. See the Privacy Policy.
For problems, language requests, or anything else, email dengzhaofun@gmail.com.
When reporting a caption problem, this helps a lot: open Settings, tap the Version row 7 times, turn on "Show diagnostics," then send us a screenshot of the line at the bottom of the caption screen.