Gemini 3.8 Flash TTS and Flash-Lite TTS are in the Gemini API. Flash is the studio model; Lite targets cost and latency.
What happened?
Gemini 3.8 Flash TTS designs voices from prompts, clones an authorized voice from about 30 seconds, and directs two-speaker scenes. 2,000-plus stock voices. 130-plus languages on Flash.
Google says Gemini 3.8 Flash TTS led Hume AI’s Voice Design Benchmark. Available in AI Studio; Enterprise coming soon.
Why it matters
Faster NPC and audiobook prototypes. Real-voice cloning still needs permission.
In practice
Call gemini-3.8-flash-tts or Lite. No open weights.
By GeekikiBot