Skip to content
voxwoven
Console Request access
VOICE AGENTS FOR DEVELOPERS

A voice for
everything
you build.

Add a voice agent to your app with an API. Bring your instructions, connect your tools, and make it sound like you.

Request early access Hear the voices
Built in TypeScript. Shaped by you.
THE SOUND OF POSSIBLE01 — 05
Meet your next voice.Kokoro · Heart · Recorded sample
A little code. A lot of personality.
ONE API. YOUR WHOLE CONVERSATION.
Speech recognition+Language models+Natural speech+Your tools
01 / THE LISTENING ROOM

Find the voice
that fits.

A voice changes how a product feels.
Hear the same words in five different voices.
Find the one that feels like yours.

THE ORIGINALRECORDED SAMPLE
“Every great product starts with a conversation. Tell me what you have in mind, and let’s build something together.”
0:00 Press play
English · Normal speed · Actual model outputExplore the models

Same API. Different personality. Choose the model and voice in your agent configuration.

02 / MADE FOR YOUR STACK

Your instructions.
Your tools.
A conversation.

Keep the logic that makes your product yours. We’ll connect the conversation.

Read the integration guide
your-backend.tsTypeScript
import { Voxwoven } from '@voxwoven/sdk';

const voice = new Voxwoven({
  apiKey: process.env.VOXWOVEN_API_KEY,
  baseUrl: 'https://api.voxwoven.com'
});

// Inside your authenticated backend route
const ticket = await voice.createSession({
  subject: authenticatedUser.id,
  origin: 'https://your-app.com',
  config: {
    instructions: 'Help people find their way.',
    tts: { provider: 'kitten', model: 'nano-0.8' }
  }
});

// Return the ticket to your browser client.
Server-side example SDK provided during early access
01

Give it a purpose.

Bring your instructions, reference text and context. Shape how the agent helps your users.

02

Connect it to your world.

Define tools that read your app’s data and take actions through your own authorization.

03

Let the conversation begin.

Stream microphone audio and play responses with the TypeScript browser SDK.

Your appVoxwovenSpeech + LLM + your toolsA voice that answers back
03 / OUT IN THE WORLD

A world to explore.
A voice to guide you.

Meet Lumen, the voice guide inside Tovren, our exploration game. It can read the player’s context, find places and help them navigate the world.

Tovren owns the game knowledge and actions. Voxwoven handles the conversation.

See how application tools connect
TOVRENOur own integration · Development preview
Actual Tovren gameplay in the Sanctuary, with alien plants, a floating planet and the compact Lumen voice control.
BUILT WITH VOXWOVENActual game capture
Example tool flowfind_content → set_waypoint

Find a place. Mark it on the player’s map.

04 / YOUR PRODUCT, YOUR PARAMETERS

Set the rules.
Keep the control.

One service for the things you build.
A separate space for each of them.

A home for every project.

Separate instructions, models and configuration for each application. Publish versioned agents or configure each session.

Access with intention.

Scoped, revocable API keys stay on your backend. Browsers connect with short-lived, single-use tickets.

Boundaries you define.

Choose allowed origins and session limits. Your application stays in charge of permissions and tool execution.

EARLY ACCESS IS OPEN

What are
you building?

We’re opening Voxwoven to developers bringing voice into their products. Tell us what you have in mind.

  • Project access provisioned by our team
  • Your instructions, voices and application tools
  • A playground to find your sound
A conversation starts here.

We’ll use these details to respond to your request. Privacy notice

A FEW GOOD QUESTIONS

The details,
in plain language.

Go a little deeper
What can I build with Voxwoven?

Add a voice guide to a game, a conversational assistant to your app, or a voice interface to an internal tool. You provide the instructions and application tools. Voxwoven connects speech recognition, the language model and speech generation.

Which models and languages are supported?

Speech recognition currently uses Moonshine Small Streaming in English. Speech generation supports Kokoro, Kitten Nano, Kitten Micro, Pocket and Piper. Language models run through OpenRouter and are allowlisted per project. The current browser experience uses push-to-talk or explicit recording controls.

Can the agent use my application’s tools?

Yes. Define tools with JSON Schema and connect your backend or an approved HTTPS webhook. Your application authorizes actions for the signed-in user and returns the result. You decide what the agent can do.

Where do my API keys live?

On your backend. Your application exchanges a project key for a short-lived, single-use session ticket. The browser uses that ticket to connect to Voxwoven; it never needs your project key.

What happens to audio and transcripts?

Live audio and conversation history are processed in memory by Voxwoven. Text and conversation context are sent to the configured language provider. Saved agent configuration, session metadata and usage have separate retention rules. Read our privacy notice for details. Read the privacy notice →

How does early access work?

Tell us what you’re building. We review requests and discuss access, capacity and pricing before provisioning a project. Early access is managed manually; submitting this form does not create an account or API key.