Repolode
Submit
kizuna-ai-lab

kizuna-ai-lab/sokuji

Real-time two-way speech translation for bilingual meetings, with cloud providers or fully offline on-device models, as a desktop app and browser extension.

Open-source alternative to:Google Meet Live TranslationMicrosoft Teams Interpreter

View on GitHubWebsite
Stars
1,408
Forks
205
Open issues
36
Last commit
2 days ago

What is it?

Sokuji translates spoken conversation in real time between two languages. It detects which language is being spoken and translates it into the other, in both directions, so two people speaking different languages can follow the same conversation. Audio can come from the microphone and system audio together, and the translated result can be played as voice or shown as live subtitles.

It ships as a desktop app for Windows, macOS and Linux that works with any app that accepts microphone input, such as Zoom, Teams or Discord. A browser extension for Chrome and Edge works directly inside web meeting platforms like Google Meet and Zoom web. Translation can run through cloud providers including OpenAI, Gemini, Soniox and Palabra.ai, or through a Local Inference mode that runs speech recognition, translation and text-to-speech on the device with WASM and WebGPU, without an API key.

Key features

  • 01Automatic detection of the spoken language with translation in both directions
  • 02Captures microphone and system audio together in one session
  • 03Live subtitles and translated voice output with a virtual microphone for other apps
  • 04Eight provider options, including cloud APIs and fully offline Local Inference
  • 05Local Inference runs ASR, translation and TTS on-device with no API key
  • 06Desktop app for Windows, macOS and Linux, plus a Chrome and Edge extension

Use cases

  • Holding a bilingual video meeting where participants speak different languages
  • Following a foreign-language webinar or game voice chat with live subtitles
  • Running translation offline on a machine without internet or a GPU, for privacy-sensitive conversations

Quick start

  1. 1

    Clone the repository

    git clone https://github.com/kizuna-ai-lab/sokuji.git
  2. 2

    Enter the folder and install dependencies

    cd sokuji && npm install
  3. 3

    Start the development version of the desktop app

    npm run electron:dev

Things to know

  • Licensed under AGPL-3.0, which requires derivative works to be open-sourced under the same terms.
  • Cloud providers need their own API keys, while fully offline mode relies on on-device models.
  • Browser extension supports only Chrome and Edge; Brave is coming soon.

Topics

#speech-translation#real-time#bilingual#speech-to-text#text-to-speech#webgpu#electron#chrome-extension#ai#linux-app

FAQ

What is sokuji?
Real-time two-way speech translation for bilingual meetings, with cloud providers or fully offline on-device models, as a desktop app and browser extension. Sokuji translates spoken conversation in real time between two languages. It detects which language is being spoken and translates it into the other, in both directions, so two people speaking different languages can follow the same conversation. Audio can come from the microphone and system audio together, and the translated result can be played as voice or shown as live subtitles.
Is sokuji free to use? What license does it have?
sokuji is open source under the AGPL-3.0 license. Check the license terms before using it in commercial products.
Is sokuji still maintained?
Actively maintained. The last commit to sokuji was on Oct 18, 2026.
What can I use sokuji for?
Holding a bilingual video meeting where participants speak different languages Following a foreign-language webinar or game voice chat with live subtitles Running translation offline on a machine without internet or a GPU, for privacy-sensitive conversations
How do I install sokuji?
Clone the repository: git clone https://github.com/kizuna-ai-lab/sokuji.git Enter the folder and install dependencies: cd sokuji && npm install Start the development version of the desktop app: npm run electron:dev
Where can I get sokuji?
The source code and installation instructions are on GitHub: https://github.com/kizuna-ai-lab/sokuji

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects