Google has launched Gemini 3.5 Live Translate, a new AI-powered audio model that translates speech in real time across more than 70 languages. Unlike conventional translation tools that wait for a speaker to finish before processing a response, Gemini 3.5 Live Translate works continuously, staying just a few seconds behind the speaker throughout a conversation, with no manual language selection required.
The model is rolling out today across the Google Translate app on Android and iOS, the Gemini Live API for developers, and Google Meet for enterprise users.
Gemini 3.5 Live Translate: What makes it different
The two biggest departures from existing translation tools are speed and voice fidelity. The model detects spoken language automatically and generates translated speech on the fly, rather than in chunks after each sentence.
More notably, it preserves the original speaker's intonation, pacing, and pitch, so the translated version sounds like the person speaking, not a generic synthetic voice reading a transcript. It is also built to handle noisy real-world environments, making it practical for settings like customer support calls, classrooms, ride-sharing pickups, and live events.
Gemini 3.5 Live Translate: How to use it
For regular users on Android and iOS, the feature is available directly inside the Google Translate app. Here's how to get started:
1. Open the Google Translate app on your Android or iOS device
2. Tap the microphone icon to enter conversation mode
3. Gemini 3.5 Live Translate will automatically detect the spoken language.
4. Speak naturally, the translated audio will play back in near real-time through your headphones.
5. On Android, a new listening mode streams the translated audio through the phone's earpiece when you hold it to your ear like a regular call, no headphones needed.
6. Any pair of headphones works for hearing the translated audio output.
Gemini 3.5 Live Translate: What enterprise users get
The Google Meet integration is a significant step up from what was previously available. The earlier version of speech translation in Meet supported only five languages. The new integration expands that to over 70 languages and unlocks more than 2,000 language pairings within a single meeting, and crucially, it no longer restricts translation to English as a pivot language.
Enterprise access through Google Meet begins this month in private preview, with a broader rollout to follow. Developers can access the model today in public preview via the Gemini Live API and Google AI Studio.
Every audio output from Gemini 3.5 Live Translate carries a SynthID watermark embedded directly in the audio waveform, identifying it as AI-generated content. Google has confirmed there is currently no way to remove it.
