Google Introduces Gemini 3.8 Flash TTS and Flash-Lite TTS with Custom Voice Design

What happened

Google DeepMind introduced two text-to-speech models on 23 September 2026: Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS. Google describes them as its most expressive audio generation models yet and says they are available across Google AI Studio, the Gemini API, Gemini Enterprise, Gemini Notebook and Google Vids.

What we know

• Flash TTS is aimed at creative direction and character design. Voices can be created from scratch with natural-language prompts that set role, accent and voice characteristics, across more than 100 languages and dialects, and each line can be directed for acting cues, pacing, dialect shifts and conversational sounds.
• Flash-Lite TTS is positioned for high-volume, cost-efficient use such as dubbing and voice agents, with control over tone and pacing.
• Google says the voice library grows from 30 original voices to more than 2,000 production-ready voices, including regional varieties such as Mexican Spanish, Quebec French and Scots English.
• Voice replication works from a 30-second audio sample of the user's own voice or a voice they have rights to, backed by consent verification, SynthID watermarking and C2PA credentials, according to the post.
• Custom voices can be saved and reused for consistent performance.
• The models join the Gemini audio family after 3.5 Live Translate, 3.5 Transcribe, 3.8 Live and 3.8 Live Extended Thinking.

What is not yet confirmed

The announcement text reviewed here does not give pricing, rate limits or latency figures, and the quality claims are Google's own. VoltKit has not generated audio with either model.

Why it matters

Voice cloning from short samples is the feature that draws the most scrutiny, and Google's answer is procedural: consent checks plus watermarking and content credentials on the output. For developers, the more practical change is directing speech line by line from an API rather than choosing among fixed presets.