Google DeepMind has launched Gemini 3.1 Flash TTS, its most controllable text-to-speech model yet. New text-based audio tags let users direct vocal style, pace, pauses and delivery. https://blossom.p...