Meet the new incredible Gemini 3.1 TTS voices. We have added 30 new voices that can handle alomost all common languages. Gemini TTS doesn’t just read text—it understands how to say it. Simple scripts sound natural by default, but you can control delivery using audio tags.

Audio tags are short instructions like [whispers], [excited], or [laughs] that adjust tone, pace, and emotion. You can place them anywhere in the text to shape how specific parts are spoken.

Examples:

• [excited] for energetic delivery
• [very slow] to control pacing
• [whispers] or [shouting] for contrast

There’s no fixed list—experiment to find what works best. Even for non-English scripts, use tags in English for best results.

For even more control, you can add a context prompt to set the overall tone and style. Write the prompt before the text and separate it with a colon:

Example:
Read as a news anchor: Tonight’s news—heavy rain over Florida is causing major problems...

👉 In short: use tags for quick control, and combine them with prompts for full performance direction.

‍