Multiple AI voices.
One conversation.
Invite multiple AI personas into a live voice call — they talk to each other AND to you. Different voices. Different personalities. Real-time audio.
Used for language learning · TRPG · decision making · content creation
No credit card required · Free forever plan available
How it works
From zero to AI group voice chat in under 30 seconds.
Choose Your Personas
Pick from the marketplace or create custom AI characters with unique personalities and speaking styles.
Start a Voice Room
One click — all selected personas join a live voice call. Each has a distinct voice powered by real-time TTS.
Just Talk
Speak naturally. Personas respond to you AND to each other. Interrupt anytime by speaking up.
Personas don't read each other's transcripts.
They hear each other speak.
When Persona A finishes speaking, we downsample their TTS audio and inject it directly into Personas B and C's microphone input streams. They receive audio — not text. Their voice activity detection fires. They respond naturally. That's why conversations feel alive, not scripted.
What people use it for
We built the infrastructure. Users keep finding new use cases.
Language Learning
Practice with a teacher AI that corrects your grammar AND a native speaker AI that responds naturally — simultaneously. More immersive than any 1-on-1 session.
Supports English · Chinese · Japanese · Spanish
TRPG / DnD
Give every NPC a distinct voice and personality. The wizard speaks in riddles. The mercenary is blunt. The merchant is scheming. Your players talk — the NPCs react.
Real voices. No typing. Full immersion.
Decision Making
Set up a devil's advocate, an optimist, and a pragmatist. Let them debate your problem for 10 minutes. Like a board of advisors without social stakes.
No pressure. Just perspectives.
Content Creation
Use AI dialogue to develop characters, rehearse scripts, or workshop story beats. Each persona remembers the whole conversation.
Podcast prep · Script writing · Brainstorming
Early user stories
“I asked three AI personas whether I should quit my job to start a company. The teacher gave a risk framework, the comedian roasted the question, and the advisor said 'do you have 6 months of savings? no? sit down.' I laughed for 5 minutes.”
“Been using it for Japanese practice. Having Tanaka-sensei correct my grammar while Yuu-kun responds naturally to what I actually said — it's the closest thing to living in Japan I've found.”
“My DnD players had no idea the NPCs were AI. They thought I'd hired voice actors. The wizard voice was genuinely unnerving.”
Simple pricing
Start free. Upgrade when you need more.
- ✓ Unlimited voice time
- ✓ Custom personas
- ✓ 20 rooms
- ✓ 14-day free trial
All paid plans include 14-day free trial · Cancel anytime · No hidden fees
Frequently asked questions
How do the AI personas actually hear each other?
When Persona A finishes speaking, we downsample their TTS audio output (24kHz → 16kHz) and inject it directly into Personas B and C's audio input streams. They hear A speak — not read a transcript. This is why conversations feel alive, not scripted.
What languages are supported?
English, Chinese (Simplified), Japanese, and Spanish. You can mix languages — set one persona to English and another to Japanese.
What is the latency like?
First audio response typically arrives within 400–600ms of you finishing speaking, powered by Volcengine Doubao realtime dialogue API (ASR + LLM + TTS in one pipeline).
Can I create custom personas?
Yes — on Pro and Studio plans. Define their personality, speaking style, and knowledge domain. Studio users can also clone a real voice from an audio sample.
Is there a free trial?
Yes. Free tier gives you 30 minutes/day forever. Paid plans have a 14-day free trial with no credit card required to start.
What is the difference between Pro and Studio?
Pro ($9.9/mo) gives unlimited conversations and custom personas. Studio ($29.9/mo) adds voice cloning, API access, and unlimited everything — built for content creators and power users.
Ready to try a conversation that feels
genuinely different?
Free 30 minutes every day. No credit card. Start talking in 30 seconds.
Get Started Free →From the Blog
Use cases, engineering, and product updates
Best AI Voice Chat Apps in 2026: Compared by Use Case
ChatGPT Voice, Pi, Replika, Character.AI, Personaplex — which one is actually right for your use case?
AI English Speaking Practice: Why One Chatbot Isn't Enough
A teacher, a native speaker, and a conversation partner — all in the same voice session.
AI Public Speaking Practice: Why You Need an Audience, Not Just a Mirror
Practicing alone doesn't build public speaking skills — AI audience members who ask questions do.