Gemini 3.1 Flash TTS is Google's text-to-speech model for expressive, steerable narration. It turns scripts into natural-sounding speech where you can direct tone, pacing, and emotion without re-recording or resynthesizing.

Why creators choose Gemini 3.1 Flash TTS

Gemini 3.1 Flash TTS brings directorial control to narration. Most TTS models read flatly. This one takes direction, and the results feel performed rather than synthetic.

Expressive, steerable narration

The model responds to inline tags like (warmly), (urgently), and (whispered) embedded in your script. You can also shape the overall tone through the prompt. A sentence marked (with doubt) reads differently than the same words without direction, and the model preserves continuity across the whole narration so the personality stays consistent from start to finish.

Natural pacing and emphasis

You can direct where the model pauses, where it speeds up, and which words to emphasize. That matters for teaching, sales, and storytelling, where a beat in the wrong place or misplaced emphasis changes what a viewer feels.

Works at any script length

Whether you're narrating a 30-second spot or a 30-minute documentary, Gemini 3.1 Flash TTS processes the whole script at once. No splicing stems together, no continuity breaks between chunks.

How to use Gemini 3.1 Flash TTS in Craze

Start a project and write your narration or bring an existing script. Craze handles the generation and integration.

1. Write or paste your script. Add inline directions like (firmly) or (thoughtfully) to shape how the model reads each part.
2. Name Gemini 3.1 Flash TTS in chat, or find it in the Models panel and generate.
3. Craze builds the audio and places it in your soundtrack track, timed to begin when the narration starts.
4. Edit the timing in the timeline, trim it to match your cuts, or regenerate a section that needs a different tone.

When you don't name a model, Craze shows compatible TTS options with their prices so you can pick before the generation runs.

Gemini 3.1 Flash TTS pricing

Pricing starts at $0.158 per 1,000 characters. The cost depends only on the character count of your final narration, not on voice style or inline directions. Use the calculator on this page to estimate the cost before you generate.

This model is available on every Craze plan, including Free. Runs draw from those credits, and if you need more, overages are billed at the same per-character rate.

Who Gemini 3.1 Flash TTS is for

Educational and instructional content

Teaching videos and tutorials need clear emphasis and natural pacing. Gemini 3.1 Flash TTS reads instructions the way a real teacher would, marking key points and pausing before a punchline.

Branded narration

Brand films and product spots rely on voice to set tone. Direct the model to match your brand voice (confident, playful, serious) and regenerate if the first pass lands differently than you imagined.

Documentary and long-form storytelling

A 20-minute piece needs consistent personality across the full runtime. Gemini 3.1 Flash TTS maintains voice and tone throughout, so your narrator sounds like a person telling a story, not a machine reading words.

Gemini 3.1 Flash TTS vs Other Models

ElevenLabs v4 is Craze's default narrator and handles multilingual scripts with accent control. Gemini 3.1 Flash TTS excels at expressive English narration when you want inline directorial control over tone and pacing. Both run on Craze at different price points. ElevenLabs v4 is the first choice for fast multilingual work; Gemini 3.1 Flash TTS is better when shaping the emotion and rhythm of a single-language narration.

For all-in-one audio generation with music and sound effects, Seed Audio 1.0 combines dialogue, foley, and ambience in one pass. If you need simple, cost-effective narration, Seed Speech from ByteDance delivers natural voice cloning without inline prompt direction. For fine-grained emotion control with tags like (angry) and (whisper), Qwen Audio 3.0 TTS offers multilingual support and two speed tiers.

Limitations and Script Requirements

On Craze, Gemini 3.1 Flash TTS has a maximum script length of 4,000 characters per generation. Longer scripts need to be split into sections and generated separately, then combined in your timeline. The model responds to inline tags and natural language direction equally well, so you can mix both styles in the same script. Audio generation takes roughly 10-30 seconds depending on script length. The model preserves voice continuity and tone consistency across the full narration, so even long scripts maintain character throughout.

Character narration often benefits from Gemini 3.1 Flash TTS over single-voice models. When you're directing a narrator to shift emotions across scenes (vulnerable in act one, confident in the climax), the inline tags give you frame-by-frame control without requesting full script rewrites. This matters for storytelling and branded content where tone progression carries meaning.

You cannot add background music, foley, or ambient sound through Gemini 3.1 Flash TTS alone. The model returns voice-only narration, so music and effects come from your project's audio track or another generation source. This makes it ideal for video projects where you're layering narration over existing music beds or sound design.