Gemini 3.8 text-to-speech says hello

Google DeepMind introduced two new text-to-speech models named Gemini 3.8 Flash and Gemini 3.8 Flash-Lite. The standard variant allows users to create unique voices using natural language prompts while controlling acting cues. The lite version focuses on high-volume needs such as dubbing and expressive agents. Both tools are available across several platforms including Google AI Studio and the Gemini API. These additions expand the existing Gemini Audio family by moving beyond static presets. Google DeepMind claims the new models secure top positions on specific industry benchmarks like Hume AI. They support over 100 languages and dialects, which helps developers build multilingual experiences. The team states this shift enables more dynamic creative control for enterprises and creators. The source does not provide verified independent data regarding user satisfaction or actual adoption rates. While Google DeepMind cites specific benchmark scores, third-party validation of these metrics is absent. The text mentions upcoming features for voice remixing but does not confirm a release date. Claims about protecting vocal talent rely on internal consent verification systems described by the publisher.