Google launches Gemini 3.8 Flash TTS voice models
ainews ·
Google has introduced two new text-to-speech models, Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS, designed for high-volume audio production and interactive scripting. The standard Flash model targets creative industries like game development and long-form narration, while the Lite version focuses on automated dubbing and customer service agents. These systems replace legacy voice catalogs with access to over 2,000 vocal profiles across more than 100 languages, including specific regional dialects. Independent benchmarks from Hume AI rank the primary model as the top performer in overall quality and accent modeling. Key features include stable audio quality over extended durations and the ability to insert non-verbal cues like laughter or sighs directly into scripts. To ensure safety, Google mandates identity verification and explicit consent for any custom voice cloning activities. All generated audio includes imperceptible SynthID watermarks and C2PA metadata to verify provenance. Early commercial integrations are already visible in platforms such as Figma, HeyGen, and Linguana.
ainews