This page may contain stale information. Last updated: 2026-09-24
Overview
Gemini 3.8 Flash TTS and Flash-Lite TTS are Google’s generative text-to-speech models announced September 23, 2026, enabling natural-language voice design, line-by-line performance direction, and consent-gated voice replication.
Key Features
- Generative voice design: Describe a voice in natural language; 2,000+ preset voices across 100+ languages
- Voice replication: 30-second sample + spoken consent verification; C2PA credentials on replicated voices
- SynthID: Inaudible watermark on all generated audio
- Tiers: Flash TTS (creative/character design) vs Flash-Lite TTS (high volume, low cost)
- Availability: Gemini API, Google AI Studio, Google Vids; Enterprise API coming soon
Google claims #1 Hume Voice Design Benchmark but benchmark author is Hume AI founder. Artificial Analysis Voice Arena ranks Flash TTS #2 (Elo 1,260).