Overview
- Google began rolling out Gemini 3.5 Live Translate Tuesday, making the feature available in the Google Translate app globally and opening public previews for developers through the Gemini Live API and Google AI Studio.
- The model processes speech as a continuous stream rather than waiting for pauses, producing translated audio that stays a few seconds behind the speaker to reduce awkward gaps and preserve intonation, pacing and pitch.
- Gemini 3.5 automatically detects more than 70 languages and supports over 2,000 language combinations in a single Google Meet session, with Meet integration starting in a private preview for select Workspace customers this month and a wider rollout planned later this year.
- Android users get a new Listening Mode that streams translations through the phone earpiece for private playback, and the system is built to handle noisy, multilingual inputs without manual language switching.
- All model-generated audio includes an imperceptible SynthID watermark for provenance, a move meant to deter misuse and help trace synthetic speech as partners such as Grab and real‑time platform vendors begin testing integrations.