Create Multi-Speaker Dialogues Directly via Natural Language Prompting
Generate complex multi-voice scripts without manual audio stitching or editing.
Use Gemini 3.1 Flash TTS to create multi-speaker dialogues by defining unique voice profiles and roles within a single text prompt.
The Scenario
You are creating a podcast episode or a training simulation that requires a back-and-forth conversation between two distinct characters.
Before & after
Previously, you had to record or generate separate files for each voice and manually stitch them together in editing software, taking 30–60 minutes per scene.
Using the Gemini 3.1 Flash TTS model, you can define 'Speaker A' and 'Speaker B' in your prompt, generating a complete, narrated scene in roughly 1–3 minutes.
The Prompt
Generate a dialogue between TWO speakers based on the following script: [PASTE SCRIPT HERE]. Use a friendly, high-pitched voice for Speaker A and a calm, authoritative voice for Speaker B. Ensure the pace is steady with natural pauses.
Gemini-TTS supports multi-speaker dialogue within a single generation flow, allowing you to assign specific voices to different parts of a script using natural language prompts.
Source
Release notes | Gemini API | Google AI for Developers"The Gemini API can transform text input into single speaker or multi-speaker audio using Gemini text-to-speech (TTS) generation capabilities."