Google has introduced two new speech generation models, Gemini 3.8 Flash TTS and Flash-Lite TTS. These tools support over 100 languages and allow users to create custom voices using text descriptions, defining specific roles, accents, and vocal traits.
The platform includes a library of more than 2,000 preset voices and a cloning feature that requires a 30-second audio sample and a recorded statement of consent. All generated audio clips contain an inaudible SynthID watermark for identification.
Users can adjust timbre, pitch, tempo, and accent through a forthcoming Voice Remixing feature. The models support script directions for dialogue, including emotional cues like laughter or sighs, and are designed to maintain voice consistency over long durations.
The models are currently available via the Gemini API, Google AI Studio, Gemini Notebook, and Google Vids. Integration is supported by developer platforms such as Agora, LiveKit, Pipecat, and Vercel, with Gemini Enterprise API access expected soon.
Pricing is based on audio output tokens, with one hour of audio equivalent to 90,000 tokens. Until the end of 2026, costs are $0.81 per hour for Flash TTS and $0.54 for Flash-Lite TTS, doubling on January 1, 2027.
