google/gemini-3-1-flash-ttstext-to-speech5 params · 8 nestedThe only model in the catalog with NO prompt field — the text lives in dialogue_turns. Filed under Music Models in the docs but it is speech, not music. Voice names are Capitalized here and lowercase on the gemini-omni-audio endpoint; the two lists are not interchangeable.