Gemini 3.8 speech models add custom voice design

Google introduced Gemini 3.8 Flash TTS and Flash-Lite TTS on 23 September. Flash TTS adds natural-language voice design, while the Lite model targets high-volume dubbing and voice applications. The release describes more than 100 languages and dialects, plus a library of over 2,000 prepared voices.

Developers can direct pacing and acting cues, save custom voices and use a thirty-second recording to reproduce a voice they own or have rights to use. Google pairs replication with consent verification, SynthID and Content Credentials. That rights condition is material: a short recording of someone else is not permission to reproduce them. Access routes include AI Studio, the Gemini API and enterprise products, with consumer uses in Gemini Notebook and Google Vids.