Google DeepMind releases Gemini 3.1 Flash TTS with Audio Tags
Google DeepMind launched Gemini 3.1 Flash TTS, adding text-command controls for vocal style across 70+ languages.
Google DeepMind has released Gemini 3.1 Flash TTS, describing it as its most controllable text-to-speech model to date. The model introduces Audio Tags — text commands that direct vocal style, delivery, and pace without manual audio editing. It is available in preview via the Gemini API and Google AI Studio for developers, rolling out in preview on Vertex AI for enterprise customers, and reaching Google Vids for general users. The model supports over 70 languages, including Hindi, Japanese, and German, and applies SynthID watermarking to all audio outputs.
For teams building voice interfaces, assistants, or content tools, Audio Tags could reduce the iteration cycle on speech output — currently a friction point when tuning delivery for specific use cases. The SynthID watermarking continues Google's push to label AI-generated audio at the model level. The announcement was made by @GoogleDeepMind on April 15.