Live voice conversation
Talk to Gemini, out loud
Press start, say something, and Gemini answers in a natural voice a moment later. It is the fastest way to practise speaking without a partner — or to hear how an AI voice handles a real back-and-forth.
Runs on Google's Gemini Live API with native audio.
Allow the microphone, then speak first — Gemini answers out loud.
- Real-time
- Voice only
- Live captions
Session type
Reply voice
Closer to a phone call than a chat window
Gemini hears your voice directly instead of a typed transcript, so pauses, hesitations and changes of direction come through the way they would with a person. When you stop speaking it replies; when you cut in, it stops.
Captions for both sides appear as you talk, which helps when you want to check a word you missed or see how you phrased something.
Hands-free turn-taking
No button to hold. Gemini notices when you finish a sentence.
Barge in any time
Talk over Gemini and it stops to listen, just like in a real conversation.
Captions as you speak
Your words and Gemini's reply appear on screen in real time.
Ten voices, four session types
Pick who answers and what kind of conversation you want before you start.
Conversations worth rehearsing
English speaking practice
Gemini keeps its language simple and, after your answer, offers a more natural way to say one thing.
Mock job interview
Name the role and Gemini asks one question at a time, with a short pointer after each answer.
Before you travel
Run through checking in, ordering a meal or asking for directions, with Gemini playing the other side.
Thinking out loud
Talk through an idea, a plan or a decision; Gemini asks follow-up questions to keep you going.
Pronunciation check
Read a sentence aloud and ask Gemini whether it understood — a quick, low-stakes test.
Hearing AI voices in action
Listen to how Gemini voices sound in conversation before choosing one for narration.
Start a session
- 1
Choose a session type
Open conversation, English, interview or travel.
- 2
Pick a voice
Ten Gemini voices, from upbeat to calm.
- 3
Allow the microphone
Your browser asks once. Headphones prevent echo.
- 4
Speak
Say hello or ask a question. There is nothing to type.
- 5
End when you're done
Press End, or the session closes at its time limit.
What happens to your voice
Your microphone audio streams from the browser to Google's Gemini Live API to produce the reply. VocaTTS does not record or store your audio or Gemini's spoken answers.
To apply daily limits and prevent abuse, a text transcript and basic usage data — start time, duration, response times — are kept. Captions disappear from the screen when you leave the page.
Guests get a few short sessions a day; signed-in accounts get more and longer ones.
Turn a conversation into content
Frequently Asked Questions
Q: Is Talk to Gemini free?
A: Yes, within daily limits. Guests can try a few short sessions; signing in gives you more sessions per day, and longer ones.
Q: Is my voice recorded?
A: No audio is stored. A text transcript and basic usage data are kept to enforce limits and prevent abuse.
Q: Can I interrupt Gemini?
A: Yes. Start speaking while it is talking and it stops to listen.
Q: Which languages can I speak?
A: Speak naturally in the language you want to practise. The English mode is tuned for English speaking practice.
Q: Why can't it hear me?
A: Check that the browser has microphone permission for vocatts.com, that you are on https, and that you use a recent Chrome, Edge, Safari or Firefox.
Q: Can I use my cloned voice in the conversation?
A: Not in live chat — it uses Gemini's built-in voices. Your cloned and designed voices work in Text to Speech, Multi-Speaker and SRT to Audio.