Gemini 3.8 TTS Ships with 2,000+ Voices, Cloning, and Open CORS Access

Google DeepMind released two new text-to-speech models, gemini-3.8-flash-tts and gemini-3.8-flash-lite-tts, on September 23, 2026, adding a library of over 2,000 voices, 30-second voice cloning, and a multi-speaker conversation API. Simon Willison's same-day bring-your-own-key playground shows the API's open CORS policy lets browser-only apps call Gemini directly, no backend proxy required.

• Google DeepMind released gemini-3.8-flash-tts and gemini-3.8-flash-lite-tts on Sept 23, 2026, with a library of over 2,000 voices.
• Custom voice cloning needs just a 30-second audio sample of a voice you have rights to use.
• The API supports multi-speaker conversations, assigning distinct voices and style instructions per character in one request.
• Caveat: the ~20-second generation time for 1m18s of audio comes from a single vibe-coded demo, not a controlled benchmark.
• Open CORS policy on the Gemini API let Willison build a client-only, bookmarkable-URL playground with no server component.

Bottom line If you're building voice interfaces or conversational agents, try the multi-speaker API and browser-direct CORS access now—it eliminates the usual backend proxy for TTS calls.