We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control
deepmind.google·6d ago
AI Engineering
DeepMind Blog·Jun 9·5 min read
Fluid, natural voice translation with Gemini 3.5 Live Translate
TL;DR
Language barriers hinder effective communication in global interactions. Gemini 3.5 Live Translate was developed to provide near real-time, natural speech translation across platforms like Google AI Studio and Google Meet.
✦ Why It Matters
Engineers can leverage Gemini 3.5 Live Translate to enhance multilingual applications and improve user experience in global markets.
Key Takeaways
- Utilize the Gemini Live API for seamless voice translation integration.
- Implement noise robustness for effective use in loud environments.
- Leverage the new listening mode for discreet translations on Android.
- Test the model in real-time applications like meetings and calls.
How It Works
Gemini 3.5 Live Translate processes audio in real-time, allowing it to translate speech continuously rather than waiting for pauses. This is achieved by balancing immediate translation with context awareness, resulting in fluid conversations that reflect the speaker's tone and pacing.