Introduction

Vaux turns speech, text or an AI agent's replies into natural speech in any voice, live. Use it to change a user's voice in real time, speak typed lines instantly, or run a briefed voice agent, all over one API.

API base URLhttps://api.vaux.ng
Realtime streamwss://vaux.ng/v1/stream
SDKnpm install @vaux/sdk

Quickstart

1. Activate the API and create a key in Dashboard → Developers. Keep the key on your server.

2. Add a token endpoint to your backend. It mints a short-lived client token for each session:

// app/api/vaux-token/route.ts (Next.js), or any server route
import { Vaux } from '@vaux/sdk/server';

const vaux = new Vaux({ apiKey: process.env.VAUX_API_KEY });

export async function POST() {
  const { token, streamUrl } = await vaux.tokens.create({ maxMinutes: 15 });
  return Response.json({ token, streamUrl });
}

3. Connect from the browser:

import { createVauxClient } from '@vaux/sdk';

const vaux = createVauxClient({
  apiKeyProvider: () => fetch('/api/vaux-token', { method: 'POST' }).then((r) => r.json()),
});

const mic = await navigator.mediaDevices.getUserMedia({ audio: true });
const session = await vaux.realtime.connect(mic, {
  voice: 'leo',
  emotion: 'excited',
  onAudioStream: (stream) => {
    audioElement.srcObject = stream; // or add to a WebRTC peer connection
  },
});

Speak into your microphone. You'll hear your words in Leo's voice.

What you can build

  • Live voice changing for calls, streams, games and social apps
  • Full identity swaps, pairing Vaux's voice with a face or character-swap video model (guide)
  • Type-to-speak for accessibility and noisy environments
  • Voice agents that answer in a chosen voice from a brief (guide)
  • Batch narration with text-to-speech

Next steps