Google DeepMind · 2026 · Gemini-native
Gemini 3.1 Flash TTS
Released April 2026. Speech treated as a language task: 200+ audio tags for style, pacing, accent, and scene direction; native multi-speaker dialogue. Google is explicit that it does not stream and is for recitation, not live agents. No reusable demo clip was obtainable for this bench.
Known for
Leaderboard-era Gemini speech with SynthID watermarking.
Sample unavailable. No redistributable clip was on file. Listen on the official site rather than a fake waveform.
- Approach
- Spoken dialogue
- Year / era
- 2026
- License class
- commercial
- Cloning
- No
- Streaming
- No
- Emotion controls
- Yes
- Languages (tags)
- en · multilingual
- Official
- Source