Google DeepMind Launches Gemini 3.5 Live Translate for Real-Time Speech Across 70+ Languages
Google DeepMind released Gemini 3.5 Live Translate, a streaming audio model that converts speech across 70+ languages while preserving tone and pitch.
Google DeepMind has released Gemini 3.5 Live Translate, a new audio model designed for real-time, cross-language speech conversion. The model supports more than 70 languages and processes audio as it streams, preserving tone, pace, and pitch to produce more natural-sounding conversations.
The model is available now in the Google Translate app and in developer preview via the API through Google AI Studio. For developers building communication tools, live customer support systems, or multilingual applications, the streaming architecture means translation can happen inline rather than in discrete chunks — a meaningful shift for latency-sensitive use cases. The pitch and tone preservation addresses a longstanding complaint about machine translation: that it strips out the prosodic cues that carry meaning and emotion in spoken language. Google DeepMind announced the release on X on June 9.