Google launches Gemini 3.8 text-to-speech models with 2,000+ voices and 30-second voice cloning

Google has launched two new text-to-speech models, Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS, turning voice generation from static presets into a dynamic creative studio. The first targets deep creative and character design, while the second focuses on high-volume, cost-effective use cases.

Users can expand from 30 original voices to over 2,000 ready-made voices across 100+ languages and dialects, including regional variants like Mexican Spanish and Quebec French. A voice can be cloned from just a 30-second audio sample, with consent verification, SynthID watermarking and C2PA credentials built in.

On Hume AI's voice design benchmark, Gemini 3.8 Flash TTS ranked first overall with 71.4 points and led in accent modeling with 60.8 points. The two models took first and second place in the overall quality index, and led competitors in Voice Arena blind tests across Japanese, Brazilian Portuguese, Vietnamese and other major languages.

Developers can try the models in Google AI Studio and Gemini API starting today. Voice cloning is unavailable in Illinois, Texas, the European Economic Area, the UK, Switzerland and India.