GitHub radar
Cicada: understand and interpret any call or livestream on your Mac
Cicada is a Mac app that shows live bilingual subtitles and writes a meeting summary while you listen to a call or stream.
It's a Mac program that listens to a call or video and instantly shows a translation and a summary of what's being said.
When it helps
Useful when you're watching a meeting, stream, or video in a language you don't speak and want to follow and keep notes in real time; not for high-stakes negotiations, since the free translation runs through Google Translate and isn't guaranteed to be accurate.
Pros
- Subtitles and translation appear instantly, no bot joins the call
- Speech recognition runs on your Mac, audio never leaves your computer
- Meeting notes link topics and action items to the exact transcript line
- The other person doesn't need to install or download anything
Cons
- Works only on Apple-silicon Macs running macOS 26
- Free Google Translate output isn't reliable for important negotiations
- The app isn't notarized by Apple, so you must approve it manually on first launch
How to set it up — step by step
- 1Open Claude Code or Codex in a folder where you want to work with Cicada
- 2Send the agent the text from the block below
- 3If macOS blocks the first launch, approve it manually in System Settings → Privacy & Security
- 4Open a call or a YouTube video and check how the live subtitles and meeting notes look
- 5Decide whether you want a paid Ling key for voice interpretation — it's optional
Text for your agent
Copy this and send it to your agent — Claude Code, Codex, any of them:
Download Cicada from https://github.com/wquguru/cicada/releases/latest, verify it against checksums.txt, and help me install it on my Mac with macOS 26 and an Apple chip (if macOS blocks the first launch, tell me how to allow it in Settings). Then open a call or a YouTube video and show me the live subtitles and meeting notes — the paid Ling API key for voice interpreting is optional, ask me if I want to add it.
You need a Mac with an Apple chip (M1 or newer) running macOS 26 — it does not run on Windows, Linux, or Intel Macs. No separate GPU is needed, speech recognition runs on the chip itself.
Open on GitHub▌ More finds