Google is rolling out two new text-to-speech models, Gemini 3.8 Flash TTS and Flash-Lite TTS, pitched at performing a script rather than simply reading it, Android Authority reports. They extend the Gemini 3.8 family that Google has been building out over recent weeks.
Flash TTS can design a new voice from a written description in over 100 languages and dialects, offers more than 2,000 ready-made voices, and can recreate a voice from a 30-second sample if the user has permission. Both models take line-by-line direction on tone, pacing, accent, dialect switches, whispers, laughs and sighs, can stage a two-person dialogue from one script and keep voices steady over hours of audio, which the outlet links to podcasts, audiobooks, dubbing and narrated video.
Google cites first place overall on Hume AI's voice design benchmark and the top two spots on its quality index, though ElevenLabs still scored higher for individual voice quality and human-like variation. Flash TTS is reaching Gemini Notebook and Flash-Lite Google Vids, with developer access through the Gemini API and AI Studio.
Voice replication requires consent verification, carries SynthID watermarks and C2PA credentials, and is blocked in AI Studio in the UK, the EEA, India, Texas and Illinois.
Sources · 1