Today, we are introducing Gemini 3.1 Flash TTS, our latest text-to-speech model with improved control, expressiveness, and quality. This enables developers, enterprises, and consumers to build next-generation AI voice applications.
Starting today, 3.1 Flash TTS is being rolled out.
Improved voice quality and control
We’ve improved the overall audio quality of Gemini 3.1 Flash TTS, making it the most natural and expressive model ever. On the Artificial Analysis TTS Leaderboard, a benchmark that captures the preferences of thousands of blind humans, 3.1 Flash TTS achieved an impressive Elo score of 1,211.
