Google today introduced two new Gemini text‑to‑speech models: gemini-3.8-flash-tts and gemini-3.8-flash-lite-tts. Both models come with a library of over 2,000 voices and allow users to create a custom voice from a 30‑second audio sample of a voice they have the rights to use.
Features and developer access
The Gemini 3.8 TTS Playground offers a bring‑your‑own‑key interface and leverages the underlying Gemini API’s open CORS policy to simplify development and integration. A prominent capability of the API is the ease of defining a full conversation between multiple characters, with each character assigned different voice and voice‑style instructions.
In one demo clip, two pelicans debate whether to move to the Pacifica Pier; the script for that clip was written by Claude 4.5 Opus, which also produced a URL to render the audio with the tool.
Performance and cost example
In a short user test, about 20 seconds were required to generate 1 minute 18 seconds of audio using gemini-3.8-flash-tts (not the lower‑cost Flash‑Lite), and the reported cost for that generation was 2.74 cents.
Why this matters
With its large voice library and support for rapid, characterized multi‑speaker dialogues, Gemini 3.8 Flash and Flash‑Lite can be applied in areas such as games, interactive storytelling, customer support, and educational voice interfaces. The simple process for creating a custom voice from a short sample and the open API access can also speed up prototyping and developer workflows.
Further details are available in Google’s documentation; this report is based on a practical test of the Playground and the example generation that used GPT‑6 Astra and Claude 4.5 Opus in the content creation chain.
