Gemini Text to Speech: Direct the Voice With a Prompt
With Gemini text to speech, you write the script and then say in plain words how it should sound, for example warm, nervous or quick. The model reads your text exactly as written and returns spoken audio you can play right away.
Cabina.AI includes Google Gemini text to speech in its list of audio models, next to ElevenLabs, OpenAI Audio and other voice tools. You don't pay for each one separately, because a single subscription lets you try all of them.
You can test Gemini text to speech free online before you choose a plan. The Free plan costs $0.00 per month and gives you 50 tokens to spend on your first voice-overs.
Voice Your Script Free on Cabina.AIWhat Gemini TTS Can Do
Style You Can Describe
Describe the sound you want, such as calm, excited or fast. In text to speech, Gemini carries that style through every line of your script, from the first word to the last.
Two Speakers in One Take
With Gemini TTS multi-speaker support, one request can voice a dialogue for up to 2 speakers. Each of them gets a separate voice, so a short scene comes back in a single take.
30 Ready Voices
Gemini TTS voices come as a set of 30 prebuilt options. Try the same lines in several of them and keep the one that suits your project.
Over 100 Languages
The newest models speak more than 100 languages. You don't need to set the language first, since they detect it from the text you type.
A Steady Voice
Your narration keeps the same pace and intonation from one sentence to the next. That steady sound helps most in long reads.
Breaths, Pauses and Laughs
Want a sigh before a punchline or a pause after a question? Mark the spot in your text, and the voice adds that moment exactly where you put it.
Pick a Voice That Fits Your Script
The Google Gemini TTS voice list has 14 female and 16 male voices. Each one has a short name, such as Zephyr, Puck, Charon, Kore or Fenrir, so it's easy to come back to a voice you liked. Beyond that set, Google describes an extended library with hundreds of extra voices on its own API.
Sample voice: Zephyr
Sample voice: Puck
Sample voice: Charon
Sample voice: Kore
How to Direct the Voice in Plain Words
Gemini speech generation takes directions in three simple ways. Put the mood, pace and tone for the whole text into one style line, such as warm and unhurried. Mark short moments like a breath or a pause right inside the script, and write a word in CAPITALS when you want it stressed. Everything else is read word for word.
Gemini TTS understands audio tags written in square brackets. Here are a few examples.
| Group | Example tags | What you hear |
|---|---|---|
| Sounds | [sigh], [laughing], [uhm] | A sigh, a laugh or a short filler sound |
| Style | [sarcasm], [robotic], [shouting], [whispering] | A new attitude or loudness for the line |
| Speed | [extremely fast] | The line read at a much quicker pace |
| Short pause | [short pause] | About 250 ms of silence |
| Medium pause | [medium pause] | About 500 ms of silence |
| Long pause | [long pause] | About 1,000 ms of silence or more |
What People Make With Gemini Voices
Video Voice-Overs and Dubbing
Turn a video script into a clean voice-over, or dub a clip for a new audience. Text to speech with Gemini is made for this kind of everyday audio work.
Podcasts and Audiobooks
Podcast episodes and audiobook chapters are long reads where clean sound matters. A Gemini voice generator can narrate both straight from a script you've already written.
Dialogues and Assistants
Write a scene for two people and hear it as a natural back-and-forth. Building a real-time assistant? Gemini text to voice also comes in a fast model made for quick replies.
How to Use Gemini TTS in Three Steps
Start on the Free Plan
Create a Cabina.AI account and stay on the Free plan for now. That way, you learn how to use Gemini TTS before you pay for anything.
Choose Gemini TTS and Add Your Script
In the list of audio models, pick Gemini TTS and type your script into the chat. If the text already exists in a document, you can paste it instead.
Set the Style and Compare
Add a style line such as calm or excited, then generate the audio. Want a different sound? Play the same script on another Gemini TTS model and listen to both.
Gemini TTS Models You Can Try
Cabina.AI lists three Gemini TTS models, and each one fits a different job, from expressive narration to quick replies in an app.
Gemini 3.8 Flash TTS
Gemini 3.8 Flash TTS is the current flagship text-to-speech model from Google. It aims for studio-grade sound and expressive acting, handles regional dialects and works in more than 130 languages. Google recommends it for two-speaker dialogue.
Gemini 3.1 Flash TTS
Google now marks Gemini 3.1 Flash TTS as a legacy preview from the previous generation and suggests moving to Gemini 3.8 Flash TTS. It voices one speaker or several and covers 87 languages, with 24 fully released and 63 still in preview.
Gemini 2.5 Flash TTS
For real-time assistants and other apps where response time matters, Gemini 2.5 Flash TTS is a fast model that's easy to control. You can steer its style and pace, and it handles one speaker or several. Google keeps serving it until further notice.
Use Gemini TTS Inside Cabina.AI
You'll find Gemini AI text to speech on the Cabina.AI list of audio models, so you open it from the same account as your other models. There's no API key to create and no code to write.
Everything happens in one chat. You can draft a script as you chat with Gemini, then turn the same text into speech with Gemini TTS. Shared context keeps your history when you switch models. You can use Gemini text to speech online, with nothing to install.
Other voice models work in the same chat. If you need a cloned voice, switch to the ElevenLabs voice generator, MiniMax Audio or CosyVoice, which all turn text into speech and can copy a voice too. For one more sound to compare, try OpenAI text-to-speech.
Simple Plans for Voice Generation
Gemini TTS pricing comes down to the plan you pick, because one subscription covers every model on the platform. Each plan adds a monthly amount of tokens to your balance.
Free
$0.00
/ month
50 Cabina Tokens
Try Gemini text to speech for free before you commit.
AI
$4.99
/ month
450 tokens
Agent
$19.99
/ month
1,000 tokens
Agent Pro
$49.99
/ month
2,100 tokens
Rather not subscribe? Pay-as-You-Go starts at $3 for 249 tokens. Larger top-ups give you 850 tokens for $10, 1,800 for $20 and 5,300 for $50.
Gemini TTS FAQ
Can Gemini Turn Text Into Speech?
Yes. Gemini can do text-to-speech with its TTS models, which take your text and read it aloud as audio. They only work in that direction, so they won't transcribe a recording or reply in a chat. For speech-to-text, the Cabina.AI audio list also has Whisper and Cabina Transcriber.
How Many Tokens Does Gemini TTS Audio Use?
Google counts Gemini TTS audio tokens per second of speech, at 25 tokens for each second. That rate applies when you call the Google API yourself. Cabina.AI tokens are a different unit, so the two numbers don't match one to one.
Does Gemini TTS Support Voice Cloning?
Google Gemini TTS does support voice cloning when you use it directly from Google, where it's called voice replication. It works only with the newest models and needs a reference recording plus a consent recording. In Cabina.AI, Gemini TTS doesn't include voice cloning.
Is Gemini TTS Audio Watermarked?
Yes. Google marks all audio from its Gemini models with SynthID. It's a watermark you can't hear, and it lets tools detect that the speech came from AI. Nothing about the sound of your voice-over changes because of it.
How Long Can the Text and the Audio Be?
A request can take up to 4,000 bytes of text and another 4,000 bytes of style prompt, with 8,000 bytes combined. One request returns up to about 655 seconds of audio, and any text past that point is cut off.
Start Your First Gemini Voice-Over in Cabina.AI
With one subscription, you get Gemini TTS and the other voice models in a single chat. Start on the Free plan, paste a short script and hear it in a voice you chose.
Hear Your Script Free on Cabina.AI