Google has added two text-to-speech models to the Gemini family: Gemini 3.8 Flash TTS for creative direction and character design, and Gemini 3.8 Flash-Lite TTS for high-volume dubbing and voice agents. Flash TTS tops Hume AI's Voice Design Benchmark and can design new voices from natural-language prompts or replicate a voice from a 30-second sample with consent verification and SynthID watermarking.
On September 23, 2026, Google introduced two new text-to-speech models to the Gemini family: Gemini 3.8 Flash TTS, built for deep creative direction and character design, and Gemini 3.8 Flash-Lite TTS, a high-volume tier optimized for dubbing, audio content creation, and expressive voice agents. Announced in a blog post by Leland Rechis and Alan Cowen on behalf of the Gemini Audio Team, the pair joins Gemini 3.5 Live Translate, 3.5 Transcribe, 3.8 Live, and 3.8 Live Extended Thinking in Google's expanding audio lineup. Both models rolled out immediately through the Gemini API and Google AI Studio, with Flash TTS coming to Gemini Notebook, Flash-Lite TTS to Google Vids, and enterprise access via Gemini Enterprise to follow.
The flagship Flash TTS turns voice creation into a promptable studio. Generative voice design lets users create bespoke voices from scratch by describing role, accent, and characteristics in natural language across more than 100 languages and dialects — Google's demos include a high-energy Melbourne DJ and a Japanese dragon. The models also ship with a library of 2,000+ production-ready voices, including regional varieties like Mexican Spanish, Quebec French, and Scots English, scaled up from 30 originals. Voice replication recreates a consistent vocal profile from a 30-second audio sample, and a voice remixing feature for fine-tuning timbre, pitch, pace, and accent is coming soon. Both models support line-by-line stage directions, scripted non-verbal cues like <laughs> and <sigh>, active-listening interjections like |mhm|, long-form generation that holds character timbre across hours of audio, and native two-speaker scene staging from a single script. Flash TTS launches with 130 languages and Flash-Lite with 101, according to SiliconANGLE.
Google positions the launch with benchmark results: Flash TTS took first place on Hume AI's Voice Design Benchmark with a score of 71.4 (including 60.8 in accent modeling), and the two models hold the top two spots on Hume AI's Overall Quality Index. In blind Voice Arena preference evaluations they topped several competitor models in Japanese, Brazilian Portuguese, Vietnamese, Modern Standard Arabic, Mexican Spanish, and Hindi. The launch intensifies pressure on specialist vendors — The Next Web notes the market already has players like ElevenLabs and Murf AI, and that Adobe added speech generation to Firefly in August. Google named Agora, LiveKit, Pipecat, and Vercel as developer platforms, and Figma, HeyGen, Wondercraft, Linguana, 99.co, and Ollang as integrating partners for dubbing, localization, and voice agents.
Safety and provenance are built into the stack: every Gemini Audio clip carries an inaudible SynthID watermark, and voice replication requires a verbal consent recording from the voice owner that matches the reference speaker, with C2PA content credentials attached to replicated voices. Google frames the release as 'transforming voice generation from static presets into a dynamic creative studio.' The combination of promptable voice design, dual-speaker screenplay control, and distribution through the Gemini API gives audiobook publishers, game studios, and voice-agent builders a mainstream alternative to purpose-built TTS platforms — while raising the bar for what studio-grade synthetic voice means.
