Report: Google opens Gemini 3.8 Flash TTS and Flash-Lite TTS

Article

Google Gemini 3.8 TTS: Voice cloning with just a 30-second sample

Beating AI News: Google has released Gemini 3.8 Flash TTS and Flash-Lite TTS. Both text-to-speech models are now available in the Gemini API and Google AI Studio. Flash focuses on audio quality, character performance, and long-text stability, while Lite is geared toward high throughput, low latency, and low cost.

Both models can directly design new voices from text, or clone the user's own voice or an authorized voice using approximately 30 seconds of reference audio. Users can control emotion, pacing, and accent line by line, and can also insert sounds such as laughter, sighs, and coughs. Two-person dialogues can be generated directly from a single script. Flash supports 130 languages, while Lite supports 101.

Pricing is also lower than the previous generation. Through December 31, 2026, Flash audio output costs USD 9 per million tokens, while Lite costs USD 6; the previous-generation Gemini 3.1 Flash TTS Preview costs USD 20. Starting January 1, 2027, the standard prices for the two models will rise to USD 18 and USD 12.

Original link https://m.theblockbeats.info/flash/368721

Key points

Google released Gemini 3.8 Flash TTS and Flash-Lite TTS, now available in the Gemini API and Google AI Studio.
Both models can clone the user's own voice or an authorized voice using approximately 30 seconds of reference audio.
Flash supports 130 languages, while Lite supports 101.
Through December 31, 2026, Flash audio output costs USD 9 per million tokens, while Lite costs USD 6.
Starting January 1, 2027, the standard prices for the two models will rise to USD 18 and USD 12.

AI-assisted interpretation

The following is analysis, separate from reported facts. Verify important claims independently.

Text-to-speech models turn written text into speech. Here, they also support designing new voices, cloning a voice from approximately 30 seconds of reference audio, and controlling emotion, pacing, accent, and sounds such as laughter.

Why it matters to readers

It combines voice design, voice cloning, and two-person script generation in a text-to-speech tool, while providing clear language-support coverage and pricing.

Beginners can think of it this way: enter text and voice instructions to generate speech with emotion, an accent, or sound effects; voice cloning requires approximately 30 seconds of reference audio.

Risks and unknowns

The article does not explain how it verifies that a reference voice has actually been authorized.
The information comes from a news brief and does not provide a Google official announcement as a supplementary source.

Report: Google opens Gemini 3.8 Flash TTS and Flash-Lite TTS

Article

AI-assisted interpretation

Why it matters to readers

Risks and unknowns

Related Developments

Related concepts

Keep following

Security Center