Gemini 3.8 Flash TTS

Gemini 3.8 Flash TTS is Google's text-to-speech model designed for deep creative direction and character voice design across gaming, immersive audiobooks, podcasts, and interactive media. Launched by Google alongside a lighter sibling variant, it is described as one of the company's most expressive audio generation models to date, reflecting continued investment in natural and controllable speech synthesis for narrative and entertainment applications.

According to the Gemini 3.8 Audio model card, the TTS variants are based on Gemini 3 Pro and accept text input up to 8K tokens while returning audio output up to 64K tokens, giving creators substantial room for long-form script-to-voice workflows. The model supports voice design and replication capabilities and, together with its Lite counterpart, ranked first and second worldwide on Hume AI's Overall Quality Index, underscoring its standing in third-party expressive-voice benchmarks for creators who need fine-grained control over tone, pacing, and character.