Agentic data science with the Data Agent Kit
cloud.google.com·6h ago
Data Science
Simon Willison's Blog·19d ago·5 min read
Gemini 3.8 TTS Playground
TL;DR
Google has launched two new Gemini text-to-speech models, gemini-3.8-flash-tts, featuring over 2,000 voices and custom voice creation from a 30-second audio sample. This playground interface allows for dynamic conversations between characters with distinct voice styles.
✦ Why It Matters
Explore the Gemini API today to enhance your applications with customizable text-to-speech capabilities.
Key Takeaways
- Experiment with over 2,000 voices in your projects.
- Create custom voices from a short audio sample.
- Utilize the Gemini API for character-driven conversations.
- Integrate GPT-6 Astra for enhanced text-to-speech functionalities.
Full Summary
Google's latest release includes two Gemini text-to-speech models, gemini-3.8-flash-tts, which offer a library of over 2,000 voices. Users can create custom voices using just a 30-second audio sample of their own voice or any voice they have rights to use.
The playground interface, developed with GPT-6 Astra, utilizes the open CORS policy of the Gemini API, enabling developers to easily implement text-to-speech functionalities. A standout feature of the API is its capability to define full conversations among multiple characters, each with unique voice styles.
For demonstration, a conversation script was generated by Claude 4.5 Opus, showcasing two pelicans debating their move to the Pacifica Pier. This innovation opens up new possibilities for interactive storytelling and character-driven applications.