Back home

Cicada

github 0.86
Visit source

Intro

Cicada (知了) is @wquguru's open-source cross-platform meeting/live-stream assistant: live bilingual captions, simultaneous interpreting and structured AI notes. Site: https://cicada.indaloop.app ; source: https://github.com/wquguru/cicada (Apache-2.0).

It listens to system audio or the mic — no bot joins the call; the other side installs nothing. Works anywhere sound plays: Zoom, Teams, Feishu, Tencent Meeting, Meet, Discord, X Spaces, YouTube, Bilibili, etc.

Platforms:

  • Mac (macOS 14.4+): fullest feature set; Apple Silicon can do on-device ASR so captions/translate/transcripts work without a key;
  • Windows 10+, Ubuntu 22.04/24.04, Android 10+: need Ling / OpenRouter / ZenMux (or compatible) AI service; iPhone still upcoming.

Interpreting: hold Right ⌥ (Right Ctrl on Windows), speak Chinese; translated speech goes into the call via a virtual mic (BlackHole / VB-Cable).

Explain

Positioning

Built-in meeting captions are often locked to one app and rarely speak for you. Cicada takes a system-audio + on-device/BYO model path and turns notes into structured, click-to-source records (conclusions, todos, fishbone charts).

Privacy

No telemetry / no first-party servers per the docs. On Apple Silicon Macs, audio stays on-device for ASR; mainly text leaves the machine (plus optional interpreting audio). Windows/Linux/Android send audio chunks to your chosen cloud ASR.

Limits

  • Not yet Apple-notarized / Windows code-signed — first launch needs a manual allow;
  • "My speech" in notes is mostly what you say while holding the talk key;
  • Non-Mac builds lean on third-party AI keys and network.

Usage

Who it's for

  • People in bilingual meetings, streams or podcasts;
  • Anyone who needs push-to-talk interpreting without a bot in the call;
  • Mac users who want OSS + on-device privacy defaults.

Getting started

  1. Download from https://cicada.indaloop.app/download/ or GitHub Releases and verify SHA-256;
  2. On Mac, grant system-audio permission in onboarding; install the virtual mic if you need interpreting;
  3. Optionally add Ling or another OpenAI-compatible service for AI notes / non-Mac ASR;
  4. Export Markdown / Word / SRT after the call, or import existing audio/video to backfill notes.

Advertisement