Sokuji
Live speech translation powered by on-device AI and cloud
59 followers
Live speech translation powered by on-device AI and cloud
59 followers
Sokuji is a cross-platform live speech translation app for desktop and browser. It supports Local Inference — on-device ASR, translation, and TTS powered by WASM and WebGPU, with no API key required, fully offline, and completely private. It also integrates with cloud providers including OpenAI, Google Gemini, Palabra.ai, Kizuna AI, Volcengine ST, Doubao AST 2.0, and OpenAI-compatible APIs.
Sokuji
Sokuji tackles a huge challenge in real-time communication - breaking down language barriers naturally and seamlessly. The approach of routing spoken translations through virtual audio devices is really clever and feels way more intuitive than just text translation.
I’m curious, how many languages does Sokuji support currently, and how does it handle accents or dialects? Also, any plans to expand beyond Google Meet integrations?
Sokuji
@evgenii_zaitsev1
Its capability depends on the AI model being used. Currently, the models provided by OpenAI support most major languages around the world, and the results are quite good.
As for accents and dialects, that also mainly depends on the AI model. We're working on integrating different models and API providers, and we believe this will improve as AI technology advances. At present, it already performs well with various English accents from around the world.
We're currently developing integration for Microsoft Teams. Zoom or a Windows client may follow.
At the moment, we only have a Linux desktop client, which functions at the system level and is not limited by specific meeting platforms—so it can be used in any scenario where a microphone is involved.
Real-time multilingual communication unlocked! 🌍🗣️ Sokuji's GPT-4o powered translation with seamless audio routing is game-changing for global teams. Finally, language barriers dissolve during video calls without awkward pauses ⚡
Sokuji
Hey everyone! Sokuji now fully supports Google Meet and Microsoft Teams, making real-time translation easier for your video meetings.
We’re also working on adding Zoom support, which will be available in v0.3.9 once it passes review next week.
If there are other online meeting platforms you’d like to see supported, feel free to comment here or reach out to us directly — we’d love to hear from you!
Tried it while practicing Spanish with a friend and it was actually fun 😄 One thing I noticed: it worked better with a headset mic than my laptop mic. Any audio hardware recommendations or best practices for clarity?
Sokuji
@hamza_afzal_butt
The main issue lies in the clarity of the input device and background noise.
Laptop microphones typically have a wide pickup range—often a fan-shaped area in front of the laptop—so they tend to capture more background noise.
This becomes particularly problematic when the monitoring feature is enabled and output through speakers, as the microphone may pick up the translated audio, leading to reduced accuracy.
However, using headphones for monitoring and a headset microphone can avoid this issue.
I don’t have a specific microphone recommendation, but here are a few things you can try:
Adjust Noise Reduction based on your environment.
In the Audio settings, you can disable the monitoring feature or avoid using speaker output for monitoring.
@jiang_zhuo Thank you!
Sokuji delivers real-time speech translation powered by GPT-4o, making multilingual communication seamless during video calls. With both a desktop app and a browser extension for Google Meet, it’s a practical and powerful tool for global collaboration.
Sokuji instantly breaks language barriers by translating speech in real time with GPT-4o, seamlessly integrating into video calls via desktop app or Google Meet extension.