Gemini 3.8 Flash TTS
Turn any script into lifelike speech with Gemini 3.8 Flash TTS — shape the emotion, stage two speakers and localize in 130 languages, free online.
Support
Pro AI Tools
Explore elite tools including the Suno AI Music Generator.
MiniMax H3
MiniMax H3 AI Video Generator
Seedance 2.5
The Future of AI Video Is Here.
Seedance 2.0
The Future of AI Video Is Here.
FLUX 3 Video Generator
Kling 3.0
Next-Gen AI Video Generator
Grok Video Generator
Create Videos from Text or Images with AI
MiniMax H3 video generator
MiniMax H3 AI Video Generator
Seedance 2.0
The Future of AI Video Is Here.
Gemini 3.8 Flash TTS: The Creative Flagship of Google's Speech Stack
When Google refreshed its speech family in late 2026, one model took on expressive acting while the other was tuned for volume at a lower cost.
- A Single Release, Two Different JobsThe flagship tier leans into acting and nuance, while its lighter sibling exists to push large batches of speech through at minimal cost.
- Steering the Performance Instead of Picking a PresetStyle notes attached to each turn, structured speech metadata and inline vocal events govern tone, tempo, feeling and accent.
- Design a Voice, or Replicate One With ConsentDescribe the voice you hear in your head using plain language, or mirror a real speaker from a reference clip paired with a matching consent recording.
How to Prompt Gemini 3.8 Flash TTS for Clean Results
Four habits that keep your transcript readable and let the performance metadata handle the acting.
What Gemini 3.8 Flash TTS Can Actually Do
From line-by-line performance control to 130-language output, here is the real capability list for Google's flagship speech model.
Expressive Performance Direction
Per-turn styling plus inline laughs, sighs, coughs, breaths and pauses — closer to coaching an actor than selecting a preset voice.
Natural-Language Voice Design
Describe an age range, personality, accent, vocal texture or role in a prompt, with more than 2,000 production voices reachable through the Voices endpoint.
Consent-Gated Voice Replication
A clean reference recording plus a matching consent recording from the same adult, wrapped in SynthID watermarking and C2PA credentials.
Two-Speaker Dialogue Staging
Write the exchange once for podcasts, teaching dialogues, product walkthroughs or game scenes — no manual line stitching required.
Long-Form Stability Without Voice Drift
Google documents steady voice identity, timbre, loudness and room tone across multi-minute narration and extended back-and-forth dialogue.
130 Languages and Regional Accents
The flagship covers 130 languages where the lighter model handles 101, plus regional accents, minority dialects and IPA pronunciation overrides.
Gemini 3.8 Flash TTS: Questions People Ask Most
Quick answers about Gemini 3.8 Flash TTS pricing, choosing between tiers, benchmark standings and the safeguards around voice replication.
What does Gemini 3.8 Flash TTS cost per audio minute?
Roughly 1.35 cents per minute of audio, based on launch pricing of $0.50 per million input tokens and $9 per million output tokens.
Which tier should I pick — Flash TTS or Flash-Lite TTS?
Choose the flagship for acting nuance and long-form narration; choose Flash-Lite TTS when you need volume and low latency.
How does it stack up against rival voice models?
Google cites 71.4 on Hume's Voice Design Benchmark, and Voice Arena ranks it second with an Elo of 1,260.
What changed from Gemini 3.1 Flash TTS Preview?
Flash-Lite TTS takes over from the 3.1 preview and drops audio output pricing from $20 to $6 per million tokens.
Can I replicate a voice, and what guardrails apply?
Replication requires a reference clip alongside a matching consent recording from the same adult speaker.
Why is the model reading my stage directions aloud?
Everything you submit is treated as a verbatim transcript, so move any lasting direction into the speech metadata field.
Run Your Own Scripts Through Gemini 3.8 Flash TTS
Try both tiers in the Gemini API or Google AI Studio — moving between Gemini 3.8 Flash TTS and Flash-Lite TTS is just a model identifier swap. Weigh batch against priority inference before you lock in a production budget.