Google has released Gemini 3.5 Live Translate, a new model designed for real-time audio translation across more than 70 languages. The system automatically detects the source language and processes speech continuously without pausing for sentence boundaries. According to the company, the output preserves the original speaker’s tone, pace, and pitch. The technology is currently accessible to developers via the Gemini Live API and Google AI Studio, while businesses can access it as a preview within Google Meet. For general consumers, the feature is available in the Google Translate app on Android and iOS devices. Ride-hailing platform Grab is reportedly testing the model to facilitate communication between drivers and passengers. All generated audio includes an inaudible SynthID watermark to identify synthetic content.
This update matters because it significantly expands the utility of real-time translation beyond the previous limitation of five languages in Google Meet. By supporting over 2,000 language combinations, the tool removes a major barrier to international collaboration and remote work. The ability to translate continuously without waiting for pauses makes conversations feel more natural and reduces friction in high-stakes or fast-paced interactions. The inclusion of tone and pitch preservation addresses a common criticism of earlier machine translation tools, which often sounded robotic or lost emotional nuance. Furthermore, the watermarking feature helps maintain transparency regarding synthetic media, a growing concern for regulators and platforms. This shift positions Google to compete more directly with competitors like Microsoft and Apple in the rapidly evolving field of conversational AI.
- Gemini 3.5 Live Translate expands support from five to over 70 languages in Google Meet, enabling 2,000+ language combinations.
- The model processes audio continuously without waiting for sentence endings and preserves the speaker’s original tone and pitch.
- All generated audio carries an inaudible SynthID watermark to distinguish synthetic speech from human voices.
Stay ahead of AI. Get the most important stories delivered to your inbox — no spam, no noise.
