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·Apr 15·5 min read
Gemini 3.1 Flash TTS: the next generation of expressive AI speech
TL;DR
AI speech generation has historically lacked expressiveness and control. Gemini 3.1 Flash TTS introduces granular audio tags for adjusting vocal style and pacing across 70 languages.
✦ Why It Matters
Engineers can leverage Gemini 3.1 Flash TTS for creating more engaging and contextually appropriate AI speech applications.
Key Takeaways
- Experiment with audio tags for precise vocal control.
- Utilize Google AI Studio for voice fine-tuning.
- Export settings for consistent voice use across projects.
- Leverage SynthID to ensure audio authenticity.
How It Works
Gemini 3.1 Flash TTS uses audio tags to allow users to control various aspects of speech generation, such as tone and pacing, by embedding commands directly into the text input. This feature enables developers to create more dynamic and contextually appropriate audio outputs.