Live Translation for Events: Using Google Meet AI

A couple of weeks ago, during the latest Google I/O, Google announced a new feature for Google Meet: live translation between English and Spanish (and vice versa). This feature is currently available for personal Google accounts subscribed to the Google One AI Premium plan.

As someone who organizes events and values accessibility, having live translation between English and Spanish is often a crucial part of our setup. However, professional interpretation services can be expensive. When I saw this new feature, I wondered if it could be used as a low-cost alternative for live interpretation during events, and the answer is yes.

In this post, I’ll explain how I set everything up, what tools I used, and what limitations you should keep in mind.

What You’ll Need

Let’s imagine you're running a live conference. You’ll need:

• An audio interface (to get the audio to your computer).
• Two personal Gmail accounts (important: Google Workspace accounts do not support this feature as of June 4, 2025).
• One Google One AI Premium subscription (only needed for one of the two accounts).
• Two computers (optional but easier than doing everything on one).
• A streaming tool like **OBS, StreamYard, **or similar.
• (Optional) A virtual audio cable tool to route audio between apps.

How It Works

Here’s a high-level overview of the setup:

Computer 1: Send the original audio to Google Meet

• Open Google Meet using the Gmail account with the AI Premium subscription.
• Start a meeting and enable the Live Translation feature.
• Select your audio interface as the input source. This should be the feed from the mic/sound console, let’s say it’s Spanish.
• Set the spoken language in Meet to Spanish.

Computer 2: Join with another Gmail account

• On a second device (or same computer with careful audio routing), join the same Google Meet with the second Gmail account (no premium required).
• You’ll see an option to enable translated audio.
• Select English as the target language, and you’ll start hearing the translated speech in real-time.
• Use a virtual audio device or OBS to capture this audio and stream it separately as the translated stream.
• Important: Make sure to mute the microphone on this second Meet participant. We only want to capture the output audio, not send anything back in.

That’s it! You’ll now have two audio streams:

• One with the original language (Spanish).
• One with the real-time translated version (English).

Caveats and Notes

• This works best with clear audio input and structured speech.
• Translation isn't perfect, but in my tests, it was surprisingly effective.
• There's a small delay, as expected with any real-time AI processing.
• You might want to monitor both streams to ensure quality during your event.
• Technical terms or very local expressions can confuse the translation engine. Sometimes it tries to translate literally or guesses the context wrong.

Example Clip

Below is a short video clip from a talk from JSConf Mexico, you’ll hear both the original and translated audio streams. It’s a short demo, but it gives a good sense of how the tool performs in a real event setting.