Arabic Desktop Audio Translator

Arabic Desktop Audio Translator

Real-time Arabic speech capture and English translation.

22 followers

Visit website
A powerful desktop application for real-time Arabic audio transcription and translation that works completely offline.
Arabic Desktop Audio Translator gallery image
Free
Launch tags:Languages•Meetings•GitHub
Launch Team
Wispr Flow: Dictation That Works Everywhere
Wispr Flow: Dictation That Works Everywhere
Stop typing. Start speaking. 4x faster.
Promoted

What do you think? …

Tanzir Rahman
Maker
📌

Some learnings from vibe coding this:

  • Start with the simplest solution first, so the code is more manageable and you can prompt the tool incrementally to match the data pipeline.

  • The Python soundcard library is the most reliable for capturing desktop audio as loopback recording, but it requires a compatible numpy version.

  • This application wouldn’t have been possible without the lightweight Helsinki-NLP/opus-mt-ar-en translation model, which is about 300 MB.

  • Latency remains an issue, and I imagine it will continue to be, as I’ve tried Samsung’s live translation feature over audio calls and found it completely unusable due to the delay.

  • You can change the translation model in the code to repurpose it for other languages.

Christy Chen

This is exactly what professionals in multi-national corps needed. Incredibly impressive work.

Congratulations on the launch!

Tanzir Rahman
Maker
@christyfea Thank you. Hoping this will help with more cross-border collaboration.
Ali Ameer

Can the app manage live transcription for lengthy meetings or is there a limit on session length to keep everything running smoothly?

Tanzir Rahman
Maker
@ali09 Lengthy meetings should not be a problem. But better memory management would definitely solve the problem if such issues arise. This tool is limited in it's accuracy but it's sufficiently helpful to better understand context than not understand anything at all. If live translations are not required, a meeting note taker would be better suited to focus on speech capture and then later translate for better accuracy.
Achraf sekri

Great product, Tanzir is using it in our team and it works like a charm