Google released two new Gemini text-to-speech models, gemini-3.8-flash-tts and gemini-3.8-flash-lite-tts, according to a post by developer Simon Willison published on 23 September 2026. Willison writes that the models come with a library of over 2,000 voices and the ability to create a custom voice from "just a 30-second audio sample of your voice or a voice you have the rights to use." Willison built a bring-your-own-key playground interface for the models, which he says was possible because of the open CORS policy of the underlying Gemini API. According to his post, a notable feature of the API is that it can define a full conversation between multiple characters, each with different voices and voice style instructions. Willison demonstrated this with a clip of two pelicans debating whether to move to the Pacifica Pier, using a script generated by Claude 4.5 Opus. He reports that generating 1 minute 18 seconds of audio took roughly 20 seconds using Gemini 3.8 Flash TTS, the higher-tier model rather than the cheaper Flash-Lite, at a cost of 2.74 cents. These figures come from a single independent source; Google's own documentation was not cited.
- Two models released: gemini-3.8-flash-tts and gemini-3.8-flash-lite-tts
- Over 2,000 voices, plus custom voice cloning from a 30-second sample
- 78 seconds of audio generated in ~20 seconds for 2.74 cents (Flash tier)
What it means for you
Google now offers text-to-speech models that can turn written scripts into spoken audio, including multi-speaker conversations and voices cloned from a short recording. The cost is very low — under three cents for over a minute of audio in one test. For most people this is a nice-to-have, not a must-have, but it lowers the cost and effort of producing narration or voiceovers.
Try this
If you produce audio content — podcasts, video voiceovers, phone greetings — try generating a short test clip and compare the quality and cost to your current method before committing to anything.
Who should care
Developers and content producers who need voiceovers, narration, or automated speech, and small businesses considering AI-generated audio for marketing or customer-facing recordings.
Skip this if
You don't produce audio content or work with the Gemini API. This is early, developer-oriented tooling with no polished consumer product yet.
Sources: Simon Willison — read the original