Gemini 3.8 TTS adds voice design and line-by-line direction

Google’s Gemini 3.8 Flash TTS and Flash-Lite TTS turn scripts into speech with controls for delivery, pacing and two-speaker conversations. Flash adds natural-language voice design, while Flash-Lite targets higher-volume workloads. Voice replication requires a matching consent recording, and the announcement says AI Studio’s replication feature is unavailable in India and several other regions.

Google · Sep 23, 2026

Why it matters

Useful for developers prototyping narration, accessibility audio or scripted voice interfaces.

More context

Try a short script through Google AI Studio or the Gemini API, checking current account access and usage pricing. Compare pronunciation and speaker consistency using your own material. Enterprise API access and voice remixing are described as coming later.

Keep in mind

These are announced capabilities, not 1min listening-test results. Replication has regional restrictions, and generated speech still needs review; watermarking does not guarantee that every misuse is prevented.

Original read