Enable Streaming Speech to Create Instant Voice AI Responses
Reduce response lag with Gemini's new streaming speech generation support.
Implement streaming for text-to-speech (TTS) using the gemini-3.1-flash-tts-preview model to create low-latency, real-time voice interactions.
The Scenario
A developer is building a real-time AI voice assistant that needs to respond instantly to user queries without waiting for the full response to generate.
Before & after
Standard TTS requires the entire text to be processed before audio playback begins, causing a noticeable delay of 5–10 seconds for long paragraphs.
By integrating TTS streaming, your application can start speaking the first words of a long response in less than 1 second.
The Prompt
Set up a streaming text-to-speech request for the following long-form content using gemini-3.1-flash-tts-preview: [INPUT_TEXT_HERE]. Ensure audio starts immediately as the content generates.
The gemini-3.1-flash-tts-preview model now supports streaming via streamGenerateContent for real-time applications.
Source
Release notes | Gemini API | Google AI for Developers"Streaming via streamGenerateContent... is now supported for the gemini-3.1-flash-tts-preview model."