What changed
Google’s November 4 Gemini API notes list Voice remixing as released. The entry gives no publication hour. [1]
Text-to-speech turns a written script into audio. In Google AI Studio or through an API, a way for software to request work, you can describe changes to a ready-made voice or one created from a description. Google saves a separate profile for new speech, leaving the original unchanged. [2]
Access and saved versions
Remixing needs a Gemini API project on Paid Tier 1 or above. The project can store up to 200 voice profiles, shared by voices created from descriptions, remixed versions, and saved voices made from recordings. Creating more variations uses that same allowance. [2]
A saved profile sets the narrator’s underlying voice; temporary delivery, such as excitement, is a separate direction. [2]
Our take
For creators and developers, start with one familiar sentence, because a pleasant voice still needs to make the important words clear. Imagine narrating a fictional workshop booking guide. Compare the original and one variation on a sentence containing the shop name, date, and booking instruction before generating the full guide.
Keep a note of which saved version belongs to the project. That gives you something repeatable to judge next time, instead of choosing the most impressive-sounding preview. We have not generated or listened to these voices.