kkdai/gemini-live-translate-macos — explained in plain English
Analysis updated 2026-05-18
Get live translated subtitles for a YouTube video or streaming audio playing in another app.
Follow along with a Zoom or Google Meet call in a different language with real-time subtitles.
Hear translated audio played back aloud alongside the original speaker's pace and tone.
Study a working example of ScreenCaptureKit audio capture combined with a Gemini Live WebSocket connection.
| kkdai/gemini-live-translate-macos | a9eelsh/heliocore | appdev/clipdock | |
|---|---|---|---|
| Stars | 13 | 13 | 13 |
| Language | Swift | Swift | Swift |
| Setup difficulty | moderate | hard | moderate |
| Complexity | 3/5 | 3/5 | 2/5 |
| Audience | developer | developer | general |
Figures from each repo's GitHub metadata at analysis time.
Requires macOS 13+, Xcode or Swift, a Gemini API key with Live Translate preview access, and granting screen recording permission for audio capture.
This is a macOS desktop app called MeetingTranslator. It listens to audio playing inside another app on your Mac, translates that audio in real time using Google's Gemini 3.5 Live Translate service, and shows the translation as floating subtitles on screen. You can also turn on a setting to hear the translated speech played back aloud. The README is written in Traditional Chinese, and the project was built shortly after Google officially released this translation API in June 2026. The key technical choice is that the app captures audio from other apps using a built-in macOS framework called ScreenCaptureKit. This means you do not need to install extra virtual audio software that many similar tools require. You pick which running app to listen to from a dropdown list, such as Chrome playing a YouTube video or a Zoom meeting window, and the app grabs only that audio. The captured audio is sent over a WebSocket connection to Google's Gemini Live API, which returns both a text transcript of what was said and a translated audio stream. The subtitles scroll in a floating window, and the translated voice plays back with minimal delay. The README notes that the translation preserves the original speaker's pace and tone. To use it you need a Mac running macOS 13 or later, Xcode or the Swift compiler to build the app, and a Gemini API key from Google AI Studio. The README includes both a one-command build script and manual Xcode setup steps. On first run, macOS will ask for screen recording permission because ScreenCaptureKit requires it even when capturing only audio. The README also documents several technical problems the author encountered while building this, including which exact model name to use in the API call and how to safely copy multi-channel audio buffers to avoid silent output.
A macOS app that listens to audio from another app, like Chrome or Zoom, and shows a live translated subtitle overlay using Google's Gemini Live Translate API.
Mainly Swift. The stack also includes Swift, SwiftUI, ScreenCaptureKit.
Use freely for any purpose, including commercial use, as long as you keep the copyright notice.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly developer.
This repo across BitVibe Labs
Verify against the repo before relying on details.