Google ships its most expressive Gemini 3.1 text-to-speech model yet with 70+ language support

Editor's take

Google has released Gemini 3.1 Flash TTS, a text-to-speech model capable of generating natural-sounding audio in over 70 languages, enhanced with granular controls for style and tone.

This advancement matters as it significantly lowers the barrier for creating localized, high-fidelity synthetic voices, impacting sectors from content creation and accessibility to global customer service. The expansion beyond basic voice cloning to nuanced emotional and stylistic expression positions it as a key competitor to existing solutions from OpenAI and Microsoft, especially for large-scale multilingual deployment.

Future developments to monitor include the model's latency and cost-effectiveness for real-time applications, its ability to handle complex linguistic nuances like sarcasm or idioms, and how quickly competitors like Amazon's Polly or Meta's Voicebox will respond with similar capabilities. The adoption rate by major content platforms will also be a critical indicator.