How to Translate a Zoom Meeting Live (Free Captions)

To follow a Zoom call in another language, open the AI Live Voice Translator in Chrome or Edge, set Listen to to Tab or screen sound, and share the Zoom tab (web client) or your entire screen with system audio (Zoom app). Spoken words appear as captions with a translation beside them as people talk. Choose Meeting: shared sound + my mic to translate both sides of the call.

Set it up step by step

1. Join the meeting — in the Zoom desktop app, or in your browser (Zoom's web client works well for this).
2. Open the AI Live Voice Translator in Chrome or Edge on your computer.
3. Under Listen to, choose:
   • Tab or screen sound — translate what others say in the meeting.
   • Meeting: shared sound + my mic — translate both the meeting and your own voice (wear headphones so your mic doesn't hear the speakers).
4. Choose the Spoken language, or Auto-detect and list the languages people may speak — for example English and Chinese.
5. Choose Translate into — your language.
6. Click Start listening. In the sharing dialog:
   • Zoom in a browser tab: pick that tab and tick Share tab audio.
   • Zoom desktop app: pick Entire screen and tick Share system audio (available in Chrome and Edge on Windows).
7. Captions appear as people speak; each phrase is translated when the speaker pauses.

Keep captions on top of the meeting

Click Floating captions to open a small window that stays above the Zoom window, so you can watch the speaker and read the translation at the same time (Chrome and Edge on a computer). Use the size buttons to make the text bigger or smaller.

Which translation engine?

• Gemini AI — the most natural translations. The recognised text is sent to iPixHub's server for translation and isn't stored; use is metered daily.
• Chrome built-in — Chrome's on-device translator (desktop Chrome 138 and later): near-instant and unlimited. Each language pair downloads once.

What runs where (privacy)

When you translate sound shared from a tab or the screen, speech recognition runs in your browser with OpenAI's Whisper model: the meeting audio never leaves your computer. Only with the Gemini engine is the recognised text sent for translation. With Whisper plus Chrome's built-in translator, nothing is sent anywhere.

The Whisper model downloads once (it's cached afterwards) and uses your graphics card when the browser supports WebGPU — expect a few seconds per phrase on an ordinary laptop.

Tips for better captions

• Good audio helps most. If people are hard to understand on the call, the captions will be too.
• Use auto-detect for multilingual meetings: each phrase is recognised and its language detected automatically.
• Wear headphones in meeting mode.
• Turn on Zoom's own captions too if your host has them enabled — the transcript can help with names and terms.

Use it responsibly

• Tell participants you're using live translation if it's appropriate — especially if you'll keep the transcript.
• Don't rely on it for medical, legal, financial or safety-critical conversations. Speech recognition and machine translation are both usually good and occasionally confidently wrong; use a human interpreter when it matters.

Frequently asked questions

Does it work on a phone?

Sharing a tab's or the screen's audio needs a desktop browser. On a phone, you can still use the microphone mode in a quiet room by placing the phone near the speaker.

Does it record the meeting?

No. It transcribes and translates as people speak; the captions can be copied, but no audio is saved. To record, use the Screen Recorder — with everyone's agreement.

Which languages are supported?

Speech recognition and translation cover dozens of major languages, including English, Chinese, Spanish, Hindi, Arabic, French, Japanese, Thai and Malay. For written documents, use the AI Translator.