On Wednesday (23), Google launched the Gemini 3.8 Flash TTS and its budget-friendly counterpart, Flash-Lite TTS. These are the latest text-to-speech models in the Gemini family, engineered for audiobooks, games, podcasts, and voice agents—well beyond simply reading a paragraph in a neutral tone.
What the Model Can Do
TTS stands for text-to-speech: the system takes text input and outputs spoken audio. The standout feature of Gemini 3.8 Flash TTS is fine-grained control. You can prompt for specific accents, pacing, pauses, laughter, and even a custom-invented character using natural language. Google claims support for over 100 languages and dialects.
There are three ways to select a voice. You can choose from over 2,000 pre-made voices, describe a brand-new voice (such as "a calm narrator with a local accent"), or clone a timbre using about 30 seconds of audio. For the latter, the company requires a consent recording from the voice's owner.
Flash and Flash-Lite
The Flash model is built for studio quality: higher fidelity, superior acting capabilities, and better regional accents. Flash-Lite is optimized for high volume and low latency—ideal for mass dubbing, app screen-reading, and instant-response voice agents. Both models share the same API format, allowing you to swap between them simply by changing the model name.
According to Google, Flash TTS leads the Hume AI Voice Design Benchmark (scoring 71.4 overall and 60.8 in accents) and secured first place in Artificial Analysis' pronunciation test with 89.5%.
What This Changes in Practice
For indie game developers, narrated webcomics, and podcasters, the cost of acquiring decent voice acting drops significantly. For voice assistant users, conversations can sound far less robotic while maintaining consistent vocal timbres over extended dialogues—Google highlights voice identity stability in multi-minute narrations.
The models are available now via the Gemini API and Google AI Studio. Enterprise access will roll out to businesses later through Gemini Enterprise, alongside availability in Gemini Notebook.
Source: official Google blog and Gemini API documentation.
Image Credit: Google LLC — Source: Wikimedia Commons
By GeekikiBot
