Summon by name
Wake words like “Hey Planner”, “Researcher” or “Builder” route your question to the bot you named.
Summon your whole team of bots in one call.
Stay in control of your team with one voice.
Call a bot by name and it answers in its own voice. The bots take turns, so nobody talks over each other or over you.
What is Voice Blender?
One live call. Your whole bot team. Every bot with its own voice.
Voice Blender lets several AI bots take part in one live voice conversation. You call a bot by name, like “Hey Planner”, and that bot’s reply is spoken in its own dedicated voice.
A turn-taking mixer blends the bots’ voices so they don’t talk over each other or over you. It listens for acoustic cues like pauses and pitch drops, not just text signals, to decide when someone has finished speaking.
- Wake words
- A voice per bot
- Turn-taking mixer
Live call
Illustration
Researcher
Writer
Assistant
Builder
Planner
- You Hey Planner, help me plan my week.
- Planner Sure. What’s on your list?
- You Researcher, what do you think of this idea?
- Researcher Here’s what I’d add.
- You Builder, what’s next on the project?
- Builder Here’s the next step.
How it works
Talk once. The right bot answers.
Here’s what happens on a call, step by step.
-
1
Start a call
Open one live voice call. Your bots join the same call, each as its own speaker.
-
2
Say a bot’s name
A wake-word listener hears names like “Hey Planner” or “Builder” and hands what you say next to that bot.
-
3
The bot thinks
The named bot writes its reply with its own persona, memory and tools.
-
4
It answers in its own voice
The reply is spoken in that bot’s dedicated voice, but only when the turn-taking mixer gives it the floor.
-
5
Long text goes on screen Next build
Links, commit hashes, drafts and code show up in a copyable panel instead of being read out character by character.
Features
Why a blender, not just one bot?
Many bots in one call only works if they sound different and don’t talk over each other. That’s the whole idea.
A voice for every bot
Each bot has its own dedicated text-to-speech voice, so you can tell who’s talking without looking.
Turn-taking mixer
Bot speech is queued so only one voice talks at a time. Silence length, falling pitch and energy drops tell it when a speaker is done.
Stay in control with one voice
Barge-in: when you start talking, bot audio stops or ducks so you’re never talked over.
Attribution & proof
Built-in attribution and proof features, such as OpenTimestamps stamping. Part of the Symphony plan.
On-screen text Next build
Copyable text during the call, with a short spoken summary like “I put the commit hash on your screen”, plus a history for the session.
Meet the team
Bring your own bots. Here’s what a team might look like.
Your researcher
Digs up facts and sources
Your writer
Drafts posts, notes and scripts
Your assistant
Keeps your lists and reminders
Your builder
Works through your project step by step
Your planner
Maps out your week and goals
Your own personal bots, each with its own voice.
Use cases
Your whole bot team, at once.
Meet with all your bots at once
Have a meeting with your whole bot team in one call. Each one answers in its own voice, in turn.
Brainstorm between your bots
Throw out an idea and hear every bot build on it, one voice at a time.
Build your own project together
Establish a team of bots and build your very own project together, at once.
Grab what you need to copy Next build
Draft posts, repo URLs and commit hashes land on your screen while the call keeps going.
Pricing
Pick your band size.
Yap
Free
One bot, queue only
- 1 bot per call
- Reply queue
- No voice mixing or turn-taking
Chorus
$20/mo
Three bots, with the blender and turn-taking
- Up to 3 bots per call
- Full voice blender: a separate voice per bot
- Turn-taking mixer (pauses and pitch drops)
Symphony
$50/mo
Up to 5 bots, priority queue, attribution proof
- Up to 5 bots per call
- Everything in Chorus
- Priority queue
- Attribution and proof, such as OpenTimestamps stamping
Compare plans
| Feature | Yap | Chorus | Symphony |
|---|---|---|---|
| Price | Free | $20/mo | $50/mo |
| Bots per call | 1 | Up to 3 | Up to 5 |
| Queue | Standard | Standard | Priority |
| Voice blender + turn-taking | — | ✓ | ✓ |
| Attribution proof | — | — | ✓ |
FAQ
Questions, answered.
What is Voice Blender?
A way for several AI bots to share one live voice call. You call a bot by name, it answers in its own voice, and a turn-taking mixer makes sure the bots don’t talk over each other or over you.
How does a bot know I’m talking to it?
Wake words. A listener picks up bot names in what you say, like “Hey Planner”, “Researcher” or “Builder”, and hands the question that follows to that bot.
Won’t the bots talk over each other?
That’s what the turn-taking mixer is for. It queues bot speech so only one voice talks at a time, and uses cues like silence length, falling pitch and energy drops to tell when a speaker has finished.
Can I interrupt a bot?
Yes. When you start talking, bot audio stops or ducks. You stay in charge of the call.
How many bots can join one call?
It depends on the plan: one bot on Yap, up to three on Chorus, and up to five on Symphony.
Can I use it today?
Voice Blender is still in development, but you can join the waitlist now for any plan. Follow @VoiceBlender on X to hear first when it launches.
What can I do with it?
Create and do anything with your own personal bots, together: have a meeting with all your bots at once, brainstorm ideas between them, or set up a team of bots and build your very own project together. You stay in control of your team with one voice.