Google Text to Speech
Google Gemini 3.1 TTS text to speech — built on Google DeepMind's latest generative speech model. Access Gemini voices and Chirp 3 HD voices through Notevibes, and direct any passage in plain words. No Google Cloud account, no API keys, no code.
Achernar
Bright & expressive · Female Google voice
Achird
Deep & assured · Male Google voice
Aoede
Warm & conversational · Female Google voice
Algenib
Crisp & energetic · Male Google voice
Every preview reads this sentence
Welcome to Notevibes. I'm here to help bring your words to life with natural, expressive speech. Whether it's a podcast, video, or audiobook — I've got you covered.
From script to finished audio
- 1
Paste your text, or drop a file
Paste straight into the chat, or drop a PDF, EPUB, Word file, PowerPoint or a link. Chapters and headings come through as they are in the file.
- 2
Noty proposes before anything runs
You see the voice Noty picked and why, how long the audio will be, what it costs in credits, and one Generate button. Change the voice or the scope first — nothing runs or bills until you tap.
- 3
Listen, direct, export
Say how a passage should sound and only that part regenerates and re-bills. Export MP3 or WAV; long documents come as one merged file per chapter.
Direct it in plain words
Natural voices are expressive out of the box. When a passage needs something else, say it in the chat — Noty writes the direction and leaves your words alone.
Per passage
A direction attaches to the passage you point at, not to the whole file.
At the beat you name
Sound moments — a sigh, a pause, a scream — land exactly where you ask for them.
One narrator, whole book
A persona line keeps a narrator consistent from the first chapter to the last.
Google Gemini TTS Prompt Guide
Gemini 3.1 TTS performs your text — it doesn't just read it. Give it three things: WHO the voice is (persona), WHAT the scene feels like (direction), and HOW a passage should land (a direction in plain words — your text stays exactly as written). Copy any recipe below as a starting point.
Recipe 01
Persona — character identity
Describe who the voice is. Age, background, vocal texture, speech habits. Gemini adapts vocal identity to match.
A weary war veteran in his late 60s. Gravelly voice from decades of smoking. Pauses before hard truths. Speaks slowly, with the weight of someone who's seen too much. Never raises his voice — but when he does, people listen.
Recipe 02
Voice Direction — scene atmosphere
Describe what the scene feels like. Location, mood, stakes. Gemini shifts pacing and energy like a voice director would.
A whispered confession in a candlelit library at midnight. Intimate, tense, afraid of being overheard. Every word carefully chosen. Long pauses between sentences — the listener needs time to absorb what was just said.
Recipe 03
Direction — say how it should sound
Say how the passage should sound in plain words, the way you would brief an actor. Noty directs it — the spoken words stay exactly as written.
Direction: Giddy launch-day energy. Burst out excited, let a laugh break through the middle, then drop to a conspiratorial whisper for the last line. Text: Oh my god, we actually shipped it! Six months of bugs and late nights and — don't tell Marcus yet, he's still fixing the staging server.
Recipe 04
Audiobook narrator — full stack
Combine persona + scene + a plain-words direction for book-length narration. The persona line keeps the narrator consistent across the whole manuscript.
Persona: A seasoned fantasy narrator in her 40s. Warm, measured, slightly theatrical without being campy. British RP accent. Comfortable with archaic dialogue and long descriptive passages. Scene: A quiet moment before battle. The calm before everything breaks. Reader must feel the weight of what's coming. Direction: Slow and low through the opening, drop to a whisper for Elara's line, then hold the tension on the last sentence. Text: The dragon stirred. Elara gripped her blade tighter. "It knows we're here." The forest held its breath.
Recipe 05
Podcast host — conversational
For podcast intros, explainers, interviews. Casual persona, natural direction, a pause where you want one.
Persona: A sharp, curious podcast host. Late 30s. Warm but direct. Thinks out loud. Comfortable with long sentences and parenthetical asides. Scene: Monday morning show open. Energy is friendly but focused — listeners are on their commute. Direction: A warm welcome, a beat before the twist, then a lift of excitement on the last line. Text: Welcome back to the show. This week, [pauses] we're doing something different. We got the interview everyone said was impossible.
Recipe 06
Commercial / ad read
For ads, trailers, promotional videos. Punchy persona, high-stakes direction, a pause where the hook lands.
Persona: A confident, polished ad voice. Mid-30s. Sounds like they've used the product themselves. Friendly authority, never salesy. Scene: 30-second spot. Hook in 3 seconds, benefit in 10, call to action at the end. Direction: Confident on the open, warm on the product name, bright and excited on the offer. Text: Everyone says sleep is the foundation. [pauses] Few products actually earn that claim. Meet Luma. Try it free for 30 nights.
Recipe 07
Character voice — non-human
Gemini handles non-human voices via a creative persona and a plain-words direction like "growl it" or "sound robotic".
Persona: An ancient dragon who speaks in a voice that rumbles like distant thunder. Words come slowly, each one chosen with the precision of something that has lived ten thousand years. A faint growl underneath every syllable. Direction: Growl the first line, go very slow through the middle, and let the last line rumble like thunder. Text: You think yourself brave, little flame. I have watched empires rise and fall. I will watch yours do the same.
Recipe 08
E-learning / explainer
For courses, tutorials, training videos. Clear persona, patient direction, a pause where the idea needs to land.
Persona: A patient senior engineer explaining a concept to a junior dev. 30s. Clear, structured, comfortable with pauses. Never condescending — genuinely enjoys teaching. Scene: Screen recording voiceover. Matching the pace of someone reading and thinking along. Direction: Clear and even, then a little warmer on the last line. Text: So the first thing to understand is that JWT tokens are stateless. [pauses] What that means in practice is — the server doesn't need to remember anything about you. Everything it needs is right there in the token.
All 30 Google voices
Tap any voice for a short preview. Every one of them takes direction in plain words.
What's inside
Gemini Persona Voices
Define WHO the voice is — "a weary war veteran", "a hyperactive 12-year-old YouTuber", "a calm NPR host". Google's Gemini model adapts vocal identity to the persona description, unlocking voices no other Google TTS product can produce.
Gemini Voice Direction
Describe the scene — "a whispered secret in a candlelit library", "a stadium hype moment". Gemini 3.1 understands context and shifts pacing, energy, and atmosphere like a voice director.
Direction in plain words
Say in the chat how a passage should sound — "whisper this", "scarier", "faster at the end" — and Noty directs it. Sighs, pauses or a scream land at the beat you name, and your words stay exactly as written.
Google Chirp 3 HD Voices
Google Cloud Text-to-Speech Chirp 3 HD voices — Google's production-grade neural voices in 30+ languages. Perfect when you need predictable, studio-clean narration instead of expressive Gemini output.
What people make with Notevibes
The most common things people make with Notevibes speech.
Most common across Notevibes
You
Reads it as lesson material, picks a clear teaching voice, and turns a timed script into a deck, one slide per block, headings never read aloud.
You
“Slower on the definitions, and pause before each example.”
Those paragraphs get the direction; the rest keeps its audio and costs nothing again.
You get
A narrated lesson as audio, or a narrated deck you can present or export.
What you get
Voices
- 2,600+ voices
- Your own cloned voice
- 30 multilingual Natural voices
Direction
- Plain words, per passage
- Sound moments at the beat you name
- Persona line for narrators
Long content
- Overnight lane, a third of the price
- Only changed paragraphs regenerate
- Price shown before anything runs
Decks & formats
- Narrated PowerPoint decks
- MP3 or WAV
- One merged file per chapter
Rights
- Commercial license on paid plans
- Downloads are yours to publish
API
- Any library or cloned voice by id
GET /api/tts/voiceslists them
Google text to speech, powered by Gemini 3.1 and Chirp 3 HD
Google’s text to speech has moved through several eras — Tacotron, WaveNet, Neural2 — each one a step closer to natural speech. The latest, Gemini 3.1 TTS from Google DeepMind, is a different kind of engine. Instead of a fixed voice reading your words, a multimodal model interprets the meaning, tone, and pacing of the text and performs it. Notevibes runs Gemini 3.1 TTS as its default expressive engine, so the Google voice you hear is the current generation, not a legacy vocoder.
The catch with Google’s own stack is that reaching those voices usually means a Google Cloud project, Vertex AI configuration, an API key, and billing setup — engineering work before you generate a single second of audio. Notevibes removes that layer. You paste your text, describe the voice, and download an MP3 or WAV. Alongside Gemini you also get Google Cloud’s Chirp 3 HD voices for clean, predictable narration across 30+ languages, so the two engines cover expressive character work and studio-neutral reads respectively.
Three layers of control, not one voice slider
Older Google TTS gives you a voice name and maybe a speaking rate. Gemini responds to direction. The first layer is persona — who the voice is: “a weary war veteran in his late 60s” or “a calm NPR host.” The second is scene direction — what the moment feels like: “a whispered confession in a candlelit library.” The third is passage-level direction: say “whisper this” or “drier, more sarcastic here” in the chat, and that passage is performed as a cue rather than a constant setting, with the words left exactly as written.
Used together, these turn a flat read into something closer to a directed take. It’s the difference between a voice that pronounces your script correctly and one that acts it.
When to reach for Chirp 3 HD instead
Not every project wants a performance. For an e-learning module, an IVR prompt, or a product demo, you often want the same clean, neutral voice from start to finish. That’s what Chirp 3 HD voices are for — Google Cloud’s production neural voices, predictable and broadcast-clean, in more than 30 languages. Notevibes exposes both engines side by side, so you can match the voice to the job instead of forcing one model to do everything.
Where Google TTS through Notevibes fits
The heaviest users are people who want Google’s voice quality without Google’s plumbing: YouTubers narrating videos, course creators, and indie authors turning manuscripts into audiobooks. The audiobook workflow leans on Gemini directly — a persona per character, plain-words direction wherever delivery shifts, and long books run overnight at a third of the price. Every render exports as MP3 or WAV with a full commercial license, so the audio is ready for YouTube, ads, courses, and client work.
Try Google text to speech free
Join thousands of creators using Notevibes for Google voiceovers — 2,600+ Google voices, 2,600+ across all languages, or clone your own.
Free to try · No credit card required
More languages
Frequently Asked Questions
What is Google Gemini text to speech?
Google Gemini 3.1 TTS is a generative text-to-speech model from Google DeepMind — the same research team behind the Gemini multimodal models. Unlike older Google text to speech products (WaveNet, Tacotron, Neural2), Gemini generates speech end-to-end with a large multimodal model, so it understands meaning, emotion, and pacing from the text itself. Notevibes uses Google Gemini 3.1 TTS as its default expressive engine.
How is Google Gemini TTS different from Google Cloud Text-to-Speech (WaveNet / Neural2)?
Google Cloud Text-to-Speech WaveNet and Neural2 are neural vocoders — you pick a voice, and they read your text. Google Gemini 3.1 TTS is a generative model — you say in plain words how a passage should sound, and the model performs the text exactly as written. Gemini handles nuance, emotion, and character voices that WaveNet cannot. Notevibes exposes both: Gemini for expressive work, Chirp 3 HD for clean neutral narration.
Do I need a Google Cloud account or Google Text-to-Speech API key?
No. Notevibes handles all Google integration for you — no Google Cloud Console, no Vertex AI setup, no Google Text-to-Speech API quota management, no billing configuration. Paste your text, pick a Gemini voice, download the audio. Zero Google Cloud engineering required.
How do I use Google Gemini voices on Notevibes?
Open Notevibes, paste your text, and pick a voice. Every Gemini voice is already expressive. To shape a passage, tell Noty how it should sound — "whisper this", "scarier", "faster and angrier at the end" — and Noty directs that passage while your words stay exactly as written. Gasps, sighs or pauses go into the text at the beat you name, and a persona line keeps a narrator consistent across long content. Generate, preview, download.
What Google voices are available?
Notevibes gives you Google Gemini 3.1 TTS voices (Aoede, Charon, Kore, Puck, and more Gemini personas), plus Google Cloud Text-to-Speech Chirp 3 HD voices across 30+ languages. That is 2,600+ total AI voices powered by Google's voice technology — more than you would get by integrating Google Cloud TTS directly.
Can I use Google Gemini TTS audio for commercial projects on YouTube, ads, and courses?
Yes. All paid Notevibes plans include a full commercial license for audio generated with Google Gemini 3.1 TTS and Google Chirp 3 HD voices. Use it in YouTube videos, YouTube Shorts, TikTok, ads, e-learning courses, audiobooks, podcasts, commercials, and client work — no royalties, no revenue share.
Is Google Gemini text to speech free?
Notevibes has a free tier to try Google Gemini 3.1 TTS voices. Paid plans unlock full book-length generation, an overnight lane for long books at a third of the price, commercial license, and priority access to Gemini capacity. You never pay Google Cloud TTS API costs directly — it is bundled into the Notevibes plan.
What audio formats and quality does Google Gemini TTS export?
Notevibes exports Google Gemini TTS and Google Chirp 3 HD audio as MP3 or WAV. Long documents come back as one merged file; chapters and paragraphs can be exported on their own. Gemini outputs are 24 kHz studio-quality — the same audio quality you'd get calling Google's model directly via Vertex AI.
Can Google Gemini TTS do character voices and audiobooks?
Yes — that is the primary use case. Notevibes' audiobook engine detects characters in your manuscript, casts a voice for each one, keeps the narrator consistent with a persona line, and directs each passage the way you describe it — with gasps, sighs or pauses placed at the beats you name. Then Google Gemini 3.1 TTS performs every paragraph as the right character. No other Google text to speech product supports this workflow out of the box.