Breaking

Google DeepMind Launches Gemini 3.8 Flash TTS and Flash-Lite TTS for Controllable Voice Generation

September 23, 2026 at 11:48 EDT

Media Generation
Foundation Models

Google DeepMind announced Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS on September 23, 2026, adding two production text-to-speech models to the Gemini Audio line. The release pushes Gemini's speech stack toward controllable voice creation, with custom vocal design, two-speaker scripts and wide multilingual coverage.

September 23, 2026 | Google DeepMind

Gemini Turns TTS Into A Voice Studio

Gemini 3.8 Flash TTS and Gemini 3.8 Flash-Lite TTS add prompted voice design, two-speaker scripts, broad language coverage and safeguards for generated speech.

130

Flash TTS languages

101

Flash-Lite languages

2,000+

Production-ready voices

30 sec

Replication sample

Current Standard Output Price

Paid tier per 1M audio tokens through Dec. 31, 2026. Each stacked block equals $1.

Flash-Lite TTS

Flash TTS

$20

3.1 TTS Preview

Build The Voice Flow

1. Describe persona

Prompt a narrator, character, accent or vocal texture.

2. Direct each line

Add speaker labels, stage cues and vocal bursts.

3. Generate audio

Render speech as WAV or streamed PCM audio.

4. Watermark and deploy

Attach detection and provenance controls.

Trust Gates Are Part Of The Product

Consent verification before voice replication
SynthID watermark on generated clips
C2PA credentials for replicated voice assets
Regional exclusions for voice replication

Creative Lift

Developers and media teams are focused on custom voices, regional accents, long-form narration, dubbing and voice agents.

Watchlist

The real test is whether the models reduce drift, voice switching, tinny long-form artifacts and uneven multilingual naturalness.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.